generic/251: fix infinite looping if fstrim_loop configuration fails

generic/251: fix infinite looping if fstrim_loop configuration fails

In generic/251, the topmost scope of the test creates a run file
($tmp.fstrim_loop), starts fstrim_loop as a background shell, and then
waits for the file to disappear.  Unfortunately, the set_*_constraints
helpers called by fstrim_loop can abort the subshell by invoking
_notrun, in which case the loop never deletes the runfile and the upper
scope waits forever.

Fix this by amending _destroy_fstrim to delete the runfile always, and
move the trap call to the top of the function with a note about why it
must always be called.

Oh but wait, there's a second runfile related bug in run_process -- if
the fstrim loop exits while run_process is looping, it'll keep looping
even though there isn't anything else going on.  Break out of the loopin
this case.

Oh but wait, there's a /third/ runfile bug -- if the fstrim_loop exits
before the run_process children, it will delete the runfile.  Then the
top level scope, in trying to empty out the runfile to get the
fstrim_loop to exit, recreates the runfile and waits forever for nobody
to delete the run file.

I hate process management in bash.

Cc: fstests@vger.kernel.org # v2024.12.01
Fixes: 2d6e7681acff1e ("generic/251: use sentinel files to kill the fstrim loop")
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>