I've been wondering, if an AP receive sensitivity is -90dBm, the CCA threshold of -82dBm doesn't apply does it? So the AP will defer transmission as long as it can hear 802.11 transmissions up to -90dBm?
No, it's a more involved algorithm.
You do mean down to -90 dBm don't you?
Part of the problem with the CCA and Energy Detect specification levels is that 802.11 hardware is getting better. Now don't get me wrong - better is a good thing.
But just for conversations sake, look at your example of -90 dBm. Several years ago only the very most expensive Enterprise gear was capable of -90 dBM sensitivity, and humdrum clients were lucky if they got the standard's -65 dBm at 54 Mbps.
Now, many AP's including SOHO versions can legitimately claim -90 or better. Some AP's get almost -100dBm for some rates ! Even mass produced 802.11n/ac client boards are getting -90 dBm. I see this all the time using an Anritsu WLAN test set.
Anyway, remember that -90 dBm is a factor of 10x weaker than -80. If the standards changed everything (e.g. energy detect levels, etc.) to higher performance levels, just because the technology got better, radios would not work effectively in older provisioned networks.
By the way, just because the standards have something specified in them does not mean that manufacturers follow the standards verbatim. Besides the intentional "holes" in the 802.11 standard, there are other incentives to break the rules. The most compelling from a technology/marketing point-of-view, is that adjusting CCA and Energy Detect levels allows for manufacturers to brew their "secret sauce". Usually people think of roaming algorithms in that regard, but that is only one of the ingredients.
Unfortunately, at least from some aspects, individuals can't change these parameters. Actual chip manufactures (Intel, Ralink, Atheros, Broadcom, etc.) may make them available to the hundreds of companies building radio devices, but before we can get hold of them these parameters have already been set. You'll never see parameters such as CCA power level listed in a radios properties page on your PC. There may in fact be "hidden" commands that can change some of them, but they are never advertised publicly.
Neither the IEEE, nor the WFA can arrest you. But the FCC can do all sorts of nasty things, so their rules are the ones that manufacturers follow most stringently.
For example, the faster the rush-to-market, the smaller the test budget, the longer the country approved list, or simply the lower manufacturing standards of a particular manufacturer, the lower the real output power that the radio will have. That also goes for radios made with the same chip, but assembled by different companies.
Each manufacturer has a target "margin". That margin is how much below the FCC power limit they are aiming for. No one dares go over- it's too risky. The same applies to FCC emissions.
The smaller the margins, the harder and more expensive it is to control parts procurement and manufacturing processes.
I realize I went a little astray, but I hope this helps answer your question.