Morning Arnulf comments inline:

I concur with Frank and Ben that the OSGeo contributor agreement already
grants Martin the right to do whatever he likes with his own submissions:

"The Foundation hereby grants the Contributor the non­exclusive,
perpetual, irrevocable, worldwide, royalty­-free, license to use, copy,
prepare derivative works of, publicly display or perform, and distribute
the Submission."

This is the only thing he is asking for right now, so we are in the clear.
Understood; he also has an outstanding request of the geotools PMC which I am going to write up
as a proposal. Should this go ahead we would be able to offer him access to more then just his direct
contributions as per his initial request on geotools-devel.
But due diligence on OSGeo's side asks for making sure that the process
does not harm any of the parties involved. Therefore I will try to
sketch some of my thoughts so that you know where I come from and in the
hope that you can correct my thinking if it went wrong anywhere.
The difficult I see, is one of mandate for this decision.  While I understand both the board
and the PMC have now been contacted, this is the "GeoTools Code Contribution Agreement" and
all parties entered into it with the understanding the GeoTools procedures would be followed with
respect to any decisions around the code base.

In short I prefer to have this handled in the same fashion as any other subcommittee.

We are in the process of putting this question to the GeoTools PMC, but as usual are limited by volunteer time.
While prior history makes this an unusual request I believe the GeoTools PMC can handle this
with integrity.

I am especially sensitive that Martin's career and livelihood is effected by these decisions.

The fork of the projects was a long and somewhat painful process for all
involved but in the end seemed to be the only way forward.

I am personally inclined to allow any project in OSGeo to choose
whatever Open Source license they want to - at more or less any time. I
am also a friend of dual or multiple licensing, given it is all FOSS
licenses.
This is one of the options to proceed with.

While we would prefer if Martin could find a home that would accept his project under an LGPL license we understand
that it is currently out of fashion and limits his options.
That said I will vote +1 for the proposed GeoTK relicensing, provided it
only includes code created by Martin (and any other individual willing
to follow along). Any code that was submitted in the faith that it will
stay under LGPL *before* the copyright assignment to OSGeo became
efective cannot be relicensed without the consent of the original creator.
It is not quite as drastic as that. The code was submitted under the stewardship
of the GeoTools PMC, as such we do have the freedom to make it available to
Martin under another license. 
So far I have not really found good arguments against proceeding like
this. If you have substantial arguments against doing this please let me
know. For me "substantial" in this context means either heavily damaging
to either project or their corresponding developer and user communities.
The course of action seems appropriate; however I think we can do more.

The relationship between the board and subcommittees is not fine (or at least not clear).  
Whether or not code has been licensed as GPL and then LGPL before or
after Martin joined and how much code he wrote and whether the fork will
be all rewritten or not is beyond my capacities to find out on my own.
This is a problem. I (and probably the rest of the board too) will have
to trust both parties on this part. I will challenge both to be
benevolent and not stingy in their estimates and opinions.
We have fairly clear headers which were subject to review during incubation
as such I do not see a problem here. 
I can see that there are concerns that a license change may have an
effect in commercial uptake. If that is the case and GeoTools wants to
also profit from these (in my opinion usually dubious and alledged)
advantages), then did you ever consider dual licensing or moving to a
more premissive version?
That is the second option - which has slightly wider implications. But yes under
that scenario we would dual license GeoTools and thus martin would be able to
start from a working subset of code.

The final option I have considered is very minimal. Martin is offering the code to
Apache on a case-by-case basis; I wish we could write a letter of understanding
with Apache and ask to release files on a case-by-case basis.
Personally I do not believe in "aggressive" licenses. Instead I find it
rather amusing that in this context permissive versions of FOSS licenses
are deemed more "aggressive" than restrictive licenses. Not long ago the
Copyleft-effect was called aggresive and "viral", interesting to see the
perception shift. Maybe my latest blog is a related read in this context:

http://arnulf.us/sevendipity/archives/50-Why-Copyleft-is-Not-a-Problem.html

In the hope of better understanding how to best proceed,
Arnulf
It is an amusing choice of language; I generally find open source licenses friendly. So
an agressive open source license is perhaps a friend who is a bit too prone to a bear hug.

Thanks for contacting us directly I am glad you are comfortable joining the email list. I am also available
as GeoTools project officer as a point of contact with the board although thus far have not been contacted
in that capacity.

I am afraid we are on the wrong side of the clock for immediate discussion. If the timezone difference
is being a problem let us know and we can try and change up who is project officer.

Jody