As a member of the steering committee and probably the one who is most cautious about pushing "swarms" allow me to respond to your well thought out post. For reference my idea of balance comes from high end competitive play, especially in timed tournament formats. I will also add that I do have experience playing swarms in this format, both as a swarm (won nationals with an ugnaught swarm '17) and against swarms (faced the Janway riot swarm aka North Carolina Swarm at nationals '18).
So we're on the same page lets define what a swarm is:
A list that utilizes a large group of low value figures that don't offer significant reward for killing. These lists often have low damage output, but make up for it by dominating objectives with high figure counts.
First, I think it's worth noting that the steering committee is committed to keeping the meta fresh and fun, which to us means a variety of archetypes should be viable. We would like to take a slow approach to achieving this, meaning we don't want to completely shake things up to a point where the meta is unrecognizable.
Another goal is increasing the number of units that are viable. There is a limit on how many units can be viable, just due to the fact that many units are very similar to each other. From that sense, the units that have the best cost/health/damage ratios will generally rise to the top.
So let's talk about regular deployments. Functionally a regular stormtrooper could be buffed to being as good as an elite stormtrooper making the only difference the number of deployments you're allowed to bring. Elites could also be buffed via point deductions to fill the role of cheap swarmy units, without allowing players to bring 4 groups.
By limiting the number of groups players can bring, lists will be less homogeneous. Elite stormtroopers could be paired with Elite jets, Elite Riots, and Regular riots to make a good swarm.
Compare that to buffing regular stormtroopers. If regular stormtroopers become the go-to swarm unit, swarm units will automatically have 4 groups. It's very very hard to balance the units to a point where it would be worth it to bring less than a full complement (either they're worth bringing and you take 4 groups, or they're not worth bringing at all).
The limit of 2 groups presents all kinds of design benefits and I think it's worth noting that FFG has also almost universally under-powered regular deployments.
Also consider that a swarm list CAN be achieved without requiring 4x of a regular deployment
With those design goals in mind we have focused on the elite deployments for now.
As far as whether swarms can be NPE or if they slow the game down here is a long rant about it. Keep in mind i'm not completely against a swarm archetype, but it's important to be conscious of how this type of archetype can potentially affect the game.
Keep this in mind-> The more figures on the board, the longer each round will take. This is especially true for units that need to take advanced positions such as riots.
The biggest slowdowns in the game are players deciding which group to activate next, and where to move each figure. For support figures like 3p0, r2, jabba, and imperial officers activations are very quick. These figures are always activated first in round 1, and their priority drops significantly in later rounds, cutting the time to decide which group to activate. These figures often don't move into contested positions so their activations are very straight forward. A riot swarm from very early in round 1 must make difficult decisions regarding the position of their figures compared to where their opponent is postured. Often these units are left exposed from as early as the 3rd activation round 1. Often the first group of riots takes as long or longer to activate as a group of weequays, or a big queen piece.
Consider that the N.C. riot swarm has 6 groups of riots to activate (and 2 officers). While round 1 for most lists flies by, the swarm player has tough choices to make for each of these activations. By the end of round 1 the riot player has spent significantly longer activating compared to the non-swarm player. Keep in mind this has nothing to do with intentionally stalling, this is just the nature of the list. Starting round 2 the riot player still takes longer to play, especially in choosing which group to activate next.
It has been claimed that since the figures are exactly the same, they should activate faster, but the opposite is true in later rounds. Since the figures are so similar, it's solely positioning that determines which group a player should activate next. This is often a very subtle and very time consuming decision. The decision of when to activate groups that vary wildly in power level is much easier than choosing which riot group to activate. (example: early round 2, do I activate Rangers first or Hera? Usually an easy choice).
Swarm players also have an incentive to slow the game down. The queen lists win by killing the swarm. The swarm list wins by getting ahead on objective points and then stalling the game until time expires. In a tournament setting with 65 minute matches, swarms will often not get out of round 3 and almost never out of round 4. Time is often called on these matches with scores in the 20-30 VP range. There is no bigger NPE than this. Again this is not achieved by intentionally stalling, but just a natural effect of the list. The riot player would need to rush to keep up the pace of play with other lists, but has no reason to rush. The non-swarm player actually needs to play as fast as possible to ensure that round 4 is reached.
To buff the swarm archetype means buffing the ability for the archetype to stall. Making swarm units cheaper increases the number of figures on the board, increasing the number of decisions/activations, slowing the game down. Increasing the health of these units makes them harder to kill, making more units left on the board in later rounds that need to activate, slowing the game down. Increasing their attack moves their playstyle closer in line with other types of lists, defeating the purpose of a "swarm" archetype.
In conclusion, I think swarms are achievable, but buffing regular deployments to the point of viability will make it harder to increase the number of viable units. Point denial strategies have the potential to be extremely frustrating, and we need to be conscious about how powerful we make units capable of achieving this strategy. I'd also like to add that if you feel that you have a good idea to balance a cheap regular deployment group like regular stormtroopers, we're listening.