Dump/Restore ext2/3/4 filesystem backup Files

Changes between versions 0.4b49 and 0.4b50 (released Mar 22, 2025)
===================================================================
1.	Fix a number of issues with static destructors in the test faketape
	helper utilities.  Fix compilation issue when HAVE_LZO was not defined.

2.	Fix reading mangled bitfields. This only affects the reading of
	historical dumps where the bitfields for the compression header were
	not in gcc order. The previous release was supposed to handle this but
	due to a bug it only worked with the gcc ordering.

3.	Minor cleanups: change a few places that used uint32_t to use
	dump_ino_t instead. Fix printf format warnings on armel. Minor changes
	for feature request #16 to get dump and restore building on musl.

4.	Don't write uninitialized data to the tape. This has no visible effect
	but prevents the tape containing random bits of memory in unused
	portions of the directory and xattr records.

5.	Always write volinfo data to the TS_END record. dump only copied the
	"tape no->first ino on tape" mapping if the archive file was created.
	Now this information is always written. (This has little practical
	effect but potentially in the future offers a way to regenerate an
	archive file without having to read the start of every tape.

6.	Use structures from e2fs lib instead of replicating them in the dump
	source. Also a small optimization that avoids the need to read in an
	inode twice to get the extended attributes.

7.	Add longopts to dump and restore. Most options now have a --long-option
	 equivalent.  Add a new --dumpdates-time option that overrides the
	"now" time calculation the dump was taken. This is particularly useful
	for dumps of snapshots where the dump time can be set to the snapshot
	time instead of several seconds later when dump actually starts.
	Add --ignore-compression-flag-for-0.4b45-dumps to allow restoring of
	corrupted "compressed" dumps written with v0.4b45. Add --filesys-name
	option to override the default "an unlisted filesystem" when what is
	being dumped is not found in either fstab or mtab

8.	Debian bug #965105 QFAfiles always had a zero tape position. Special
	thanks to Alexander Zangerl for noticing that I'd inadvertently dropped
	the debian patch when building a trial debian package for v0.4b49
	instead of including it in the v0.4b49 release of dump.

9.	Fix archive file handling - the archive file was not padded to a
	multiple of ntrec records, instead it contained the same number of
	TS_END records as the tape. This lead to a regression in v0.4b49 which
	could not read a QFAfile unless it happened to be the right length.