[PATCH] xfstests: enable generic filesystems on some ENOSPC tests

[Christoph Hellwig]

The change to 204 and the infrastructure seems fine for me, but I don't
really like how the other use an explicit agcount (which at least for
109 was very intentional) for xfs and just defaults for others.  At
least 109 should be duplicated into a generic one run everywhere and
an xfs-specific with the agcount.  Need to look a bit more at the
history of 083 to max a decision for that one.  And some less cryptic
variable names in the new code would also be nice.