Great thoughts Carlos...
The way I typically look at it Carlos is to look at a few parameters, namely:
1. the total bps of the given radio first.
In the case of AP-105 you have 300Mbps max, AP-115 you have 450Mbps max, the AP-135 450 Mbps. So right away, the AP115 and AP135 have advantages of 50% over the AP-105. That is... as long as... your clients are 3 stream 802.11n clients. If they aren't then they don't 'see' the 50% throughput capability improvement and you have apples:apples from a throughput standpoint.
2. the vintage of the access point.
We have hundreds of people in R&D every hour of every working day. Those add up in my opinion. A simple look at the birthdays will tell you the AP-105 is 2009 vintage, the AP-135 last year, and the AP-115 this year as examples.
3. the memory and cpu design
In the case of the AP-135, it has the largest allocation of 'resources' of the APs I am discussing in this reply. Thus, it stands to reason it can be loaded to the highest levels regardless of the capabilities of the clients attaching.
Said another way:
When I want maximum density; go AP-135 :) (Or better yet AP-225 !!)
When I want 3 stream 450Mbps 802.11n, medium density, go AP-115
Everything else, I continue to go AP-105
JF
PS - there is no exact number on client count, as that is but one parameter to consider. Think of 50 iPhones sitting in pockets, vs. 10 iPads downloading the recent iOS7 update. Are the 50 iphones more or less of a loading on the network than the 10iPads? Pretty easy to see there that we need to take into account utilization.
PPS - my record on an access point? 220 clients from a K-12 school. Routinely I see APs pushing 50,100,150,200 clients. They all work, however everyone 'waits their turn' as you point out. Far from ideal. The solution? More radios, and our client match in those high usage areas.