You aren't overthinking it - you are thinking critically about it <--this is a good thing.
It looks like you've done plenty of research and know what you're getting into. You also recognize the need for good pitching rates to produce the best beer possible. These are all good things.
To get to your questions, you are correct, reusing yeast IS a lot of guesswork on many fronts: viability of yeast, vitality of yeast, quantity of yeast per milliliter. Without having the tools needed to perform cell counts and vitality testing, you are at the mercy of your knowledge and the universe. Luckily, there are some generalizations you can use to help you along your way.
Yeast substrate (Yeast slurry and washed yeast) will have anywhere from ~1-8 billion cells per milliliter. The thinner the substrate the less the count. The thicker, the higher the count. You will likely never achieve the 5-8 billion cells per milliliter in your non-professional setting - reserve those counts for commercial liquid yeast (i.e. white labs, wyeast). However, you can get from 1-5ish at home. I would reserve the 3-5 counts for multi-step starters when starting from small quantity of yeast and using nutrients and O2. So now we're down to 1-3 billion cells per milliliter for yeast that has been collected post-fermentation. I usually pick a number based on thickness of the yeast substrate I will be using. When in doubt, pick lower.
Yeast will also contain a portion of non-yeast material that can range from negligible to significant. You will have to rely on your own observations to determine an approximate percentage of non-yeast material in your yeast substrate. Using Mr Malty as a guide on this aspect you can assume about 10% is non-yeast UNLESS you can visibly see plenty of non-yeast material in there, in which case you're better off to estimate a up to 25%. When I collect directly from the yeast cake without rinsing, I estimate at least 25%. When I was rinsing my yeast really well I would estimate 10% or even less sometimes. It largely comes down to how well you think you've done as separating the non-yeast material from the yeast material in your washing process. Obviously, you can't distinguish between dead and live cells but if you've done a great washing job then assume down to 10%. If you've simply done a quick rinse then aim around 15-20%, and if you've simply collected yeastcake then aim for 25-30%.
So now you know how you can estimate the approximate amount of yeast in a milliliter of yeast substrate (cells/milliliter minus percentage non-yeast material). This is a pretty good way of estimating how big of a starter to use and how much yeast to use in your starter to determine how much propagation will occur using a tool like (yeastcalc: yeastcalculator.com). The only thing you don't know is the viability of the yeast you've collected and stored. This, again, is a guessing game based on the beer the yeast fermented, how high a gravity the yeast went through, how long it sat under the beer at warm temperatures, how long it sat after being collected, if it was rinsed/washed or just stored directly from the cake, and a number of other factors. I generally will only keep yeast from low-ish gravity beers (1.050 or less) and like to collect my yeast fairly quickly after fermentation (within 7-10 days). I will pitch directly from my collection container if it's been within about 3 weeks. If it's been longer, say 3-6 weeks, I'll pull a medium size sample from the collection container and build a single step starter. If it's been a lot longer, 6-12 weeks then I'll pull a small size sample from the collection container and build a multi step starter. I always use O2 in my starters, and will alway use nutrients in my really old yeast starts or frozen yeast starts.
I've probably overlooked some aspects of my process or estimating procedures, but I think this gives a good overview of how I proceed with reusing yeast. I hope this helps.
In you specific case, this is how I would proceed without knowing much about your collection/washing process:
-You collected yeast from a perfect gravity beer.
-A starter is a good idea for 3-4 week old yeast
----I have occasionally pitched from 3-4 yeast old yeast but you may as well try out your stir plate
-Yes, your starter size is largely determine by the gravity and volume of the beer you'll be brewing
-Don't fret too much on over-/under-pitching, but it's generally best to over-pitch if you have to choose one.
-Don't reduce your starter to match your large quantity of yeast; reduce the amount of yeast you put in your starter.
-Given the initial beer gravity, and storage time, I would not be concerned with viability unless that beer sat for many weeks before collecting the yeast.
-The formula you found, I like. Many folks subscribe to the 1million/millilieter/°P, but I like 0.75 better
-That formula assume you'll be making a 5.5 gallon batch; adjust accordingly (21,000 ~= 5.5 gallons)
-Use the formula to determine how many yeast cells you'll need
-Then use yeastcalculator.com to determine how big a starter you'll need to produce your final count of yeast giving the appropriate starter gravity (1.030-1.040), and innoculation rate.
-I would personally aim to use about 75 ml of your yeast slurry in your starter (decant top liquid first)
-I would assume 2bill/milliliter (~150 billion cells) with an 85-90% viablity (~125 billion cells)
-I would use stirplate option on yeastcalc with K Troester, but J Zainasheff is fine also.
-I suspect you'll only be looking at needing an ~1 liter starter.
-Give yourself time to finish the starter and crash cool it for 1-2 days prior to brewing.
Again, Hope this helps. (my fingers are tired
)