[patch, fortran] Fix PR 26769: transpose and reshape for real(kind=10), revert automake to 1.9.3

[FX Coudert]

:REVIEWMAIL:

>> 2006-03-23  Thomas Koenig  <Thomas.Koenig@online.de>
>>
>> 	PR fortran/26769
>> 	* iresolve.c (gfc_resolve_reshape):  Remove doubling of
>> 	kind for complex. For real(kind=10), call reshape_r10.
>> 	(gfc_resolve_transpose):  For real(kind=10), call
>> 	transpose_r10.

This approach of the problem (using same lib functions for all types) 
has been annoying me for some time.

What happens on machines with a real(kind=16) but no integer(kind=16)? 
On when we (though, to my knowledge, this second case doesn't exist 
today) have a real(kind=10) and integer(kind=10) with different paddings 
or other stuff?

I'd really think it would be much better having different routines for 
real and integers. The cost for that isn't worth the price of worrying 
about this kind of assumption every time we have a new real or integer 
kind available, in my opinion.

Other than that, the patch is technically OK. So, while I think this 
patch shouldn't be applied, if someone wants to tell me that I suck, the 
patch will then be pre-approved! :)

FX