We're making changes to the Gecko toolkit script which can have such an impact on Thunderbird access, and we're counting primarily on the functional testing which I do and which Mike does to catch any and all regressions.  And one of us (okay, me) is an Evo person. :-) Therefore, we need systematic, repeatable regression tests to ensure that changes intended for Firefox are not negatively impacting Thunderbird.

We can probably do something clever with profiles, perhaps even make an "orca" gmail account to which we can forward problematic messages, etc., etc. 

I'll look into this and implement it.  But input is most certainly welcome.  Thanks!