fortran: fix off-by-one string copy error

[jsquyres]

Followup to commit 694e78a (i.e., PR #13381): Ben Menadue @benmenadue correctly pointed out that < should have been <=.

Refs #13375.

[bosilca]

Have you tested it with the reproduced provided on the initial issue ?

[hppritcha]

sorry didn't catch this last time!

[jsquyres]

Have you tested it with the reproduced provided on the initial issue ?

Yes -- tried all 3 cases:

1. Original c string is 0123456789abcdef (i.e., 16 characters + \0)
2. The "f string" is actually 32 characters long, and is memset zeroed out before each of the 3 cases (so that I can opal_output the result)
3. case 1: call ompi_fortran_string_c2f() with len=8.
4. case 2: call ompi_fortran_string_c2f() with len=16.
5. case 3: call ompi_fortran_string_c2f() with len=20.

[Little-Beezle.local:85670] JMS: cstr = 0123456789abcdef
[Little-Beezle.local:85670] JMS: strlen(cstr) = 16, len = 8
[Little-Beezle.local:85670] JMS: Called with fstr len=8, result: -->01234567<--
[Little-Beezle.local:85670] JMS: strlen(cstr) = 16, len = 16
[Little-Beezle.local:85670] JMS: Called with fstr len=16, result: -->0123456789abcdef<--
[Little-Beezle.local:85670] JMS: strlen(cstr) = 16, len = 20
[Little-Beezle.local:85670] JMS: Called with fstr len=20, result: -->0123456789abcdef    <--

[jsquyres]

Did a no-change-force-push in an attempt to clear the error with community / AWS CI (there was a disk space problem on the CI nodes that has now been fixed).