incremental rebuilds after icu bump

[Caolán McNamara]

On Fri, 2012-05-11 at 13:13 -0700, julien2412 wrote:
> Just to say, after having read some very good advice from Fridrich on
> #libreoffice-dev that after several tests, this worked for me :
>  make icu.clean && make i18npools.clean
>  make icu && make icu.deliver
>  and finally make i18npools
> 
> Perhaps it may help someone else.

That looks better now that I think about it, though
s/i18npools/i18npool/

FWIW, the problem likely goes away if 
a) the icu makefile.mk was changed use the same OUT2INC pattern as the
rest of the those 3rd party modules do. copy, without preserving
timestamps, the headers from location module/untarred-somewhere/X to
module/inc/X at the end of the custom build and deliver from there to
solver.
b) generally touch everything in a 3rd party module after untarring,
with the probable nasty side-effect of autotools kicking in to
regenerate Makefile.in from Makefile.am, etc. and having a fit.

On the positive side of things, the bright note is that because the
dependency stuff in gbuilded module is working so well for so long that
its a such a WTF when something happens that subverts it rather than the
daily event it used to be.

C.