gc: Consider ckmeta/ckdata successful even if we mutated the filesystem

Also moved ckmeta/ckdata progress into lfs->flags. We have the bits
available so we might as well use them instead of allocating bools on
the stack...

Whether or not to consider ckmeta/ckdata successful when the filesystem
has been mutated is a bit nuanced.

Initially, I thought we trigger a re-traversal, since we may have
introduced new blocks that haven't been checked. But think about it,
where did those blocks come from?

Any new blocks introduced by filesystem mutation will have just been
written. And if a write introduces corruption you probably have bigger
problems...

... Actually as I write this I realized mounting without ckprogs makes
this even more nuanced, but since ckmeta/ckdata is more intended for
data-at-rest error detection I'm going to keep the change for now.

If you want to catch write errors, you really should enable ckprogs.
This is only a problem for lfsr_fs_gc, and the use cases for
ckmeta/ckdata in lfsr_fs_gc will probably catch any write errors on the
next cycle anyways...

Code changes:

           code          stack
  before: 36208           2680
  after:  36244 (+0.1%)   2680 (+0.0%)

lfs.h

@@ -140,6 +140,8 @@ enum lfs_fsinfo_flags {
 
     // internally used flags
     LFS_F_ORPHANS       = 0x8000, // Filesystem may have untracked orphans
+    LFS_F_CKEDMETA      = 0x2000, // Filesystem metadata checked during gc
+    LFS_F_CKEDDATA      = 0x4000, // Filesystem data checked during gc
 };
 
 // File types