olivier,
concerning the issue raised of the pattern field disappearing, you should change your ticket to a bug. I’ve been struggling if I should change that behavior or not (it’s a symptom of the jquery plugin we are using but I believe I can change the behavior of it). I had not yet decided if it should be ‘fixed’ or not, waiting to see if others brought it up.
concerning the custom context, I will first address mickecarlsson. Mick, I believe that custom context is very much a business feature probably more so then a home/hobby feature. It has a lot of issues which are unknown to a lot of people who use it to create a sense of multi-tenancy like olivier mentions in one of his cases. It’s fine if you understand the ‘security’ implications though I suspect most who use it that way do not. But it does provide some very useful features that can not be done elsewhere.
This module has never been part of core for a variety of reasons one of them being that we knew the day was coming that the outbound route schema structure was going to change and break it. There has been much warning and have also been people who have stepped up to the plate like kenn10 to spear head a drive to provide the funds for someone to update it for 2.8. I’ve already contacted naftali5 but he is just too swamped to continue to maintain it.
The fact of the matter is you are right that there are a lot of businesses and in particular resellers who use this module. As a result, and if it is really as important as you and many say it is, then there should be no issue for you and all the others who are profiting from it to spend a few hundred dollars each when the collection pot comes around to pay for someone to make sure it is up and running in 2.8. (I’ll be looking for your name to spear head the drive and throw in the first donation into that pot 
Of course if one of these groups who makes heavy use of it wants to simply adopt the code and make the changes necessary that would be just as valuable. Remember, the purpose of Open Source is a bi-directional street, to give back as much as to consume.
Concerning Advanced Routing, I am not aware of which module you are referring to either. There are a handful of modules that handle outbound routing. Route permissions is in the contributed svn repository but not in the online extended repository as there were issues that are broken from the start that have not been addressed. If it is something else that is in general use, then it would be wise of the author to have a look at upgrading the module, and even wiser to toss it into our repository for more developer eyes to have access to.
We make extraordinary efforts to maintain compatibility with new versions and do better than most projects in doing such. However, there are some things that do require change and this is one of them. It is also why we have gotten 2.8 out as early as we have and why we are leaving it for a very long cycle to both catch issues and give other developers an opportunity to check their modules and make changes that may be necessary.
As far as where you think efforts should be spent on 3.0 vs 2.x you are clearly entitled to your opinion. However I get the impression that your opinion is that we are going to finally have the opportunity to use FreeSWITCH in place of Asterisk and everything is going to get better. That is the message I have gotten from other posts you have provided. I think you need to get real about this perspective. There are a lot of nice things that FreeSWICH brings to the table and for all I know, in your specific situation maybe you have already evaluated it against your own needs and find that it is a much better match. However, it comes with its own baggage and will have the same type of growing pains as Asterisk and any complex project has. As an example, I’ve personally appreciated that I don’t have to restart Asterisk to fix memory leaks any longer. I’m sure there are still ones hiding in there or new ones being introduced, but its maturity has made this a thing of the past. I can’t say the same for FreeSWITCH but I’m sure over time they will also find their issues.
However, the other reality is that Asterisk is much more feature rich and from a PBX perspective, especially geared at the SMB market as FreePBX is, I think you will find that Asterisk may likely mature much quicker on FreePBX v3 then FreeSWITCH. I personally hope that we help in some small way for both products to further flourish but as it stands today, Asterisk still has a few years of head start on features that FreeSWITCH will need to catch up on, and Asterisk is not standing still either.
So in conclusion, you are clearly welcome to your opinions and even more importantly, maybe you want to get involved and help push v3 along faster through some development effort (or find a way to sponsor a developer…). In the meantime, there are hundreds of thousands of 2.X systems out there, many of them depending on features that are going to take time to mature. We are not going to turn around and abandon that installed base by prematurely pulling resources from 2.X. Furthermore, it’s Open Source and not something that “we” can do as “we” don’t control the open source community. You will notice that a substantial amount of the development that has not only gone into 2.7 (and prior releases) but is also going into 2.8 comes from the community and that is not going to go away any time soon. In the same way, v3 has made some amazing progress for the same reason, because it has substantial community support and some great developers who have gotten interested and involved. I personally hope that once we get the general preview release out we can stimulate more interest for more contributions to help v3 quickly catchup to v2 so we can eventually do what you suggest, once it is at a point where the bulk of the critical features and capabilities are present to allow much wider migration.