Use f_bavail to query available space. Not f_bfree.

f_bfree space might not be usable. In practice we see failures in big
disks because f_bfree is over 2GB and f_bavail is 0. We get ENOSPC if
writing on disk then.

buildstream/_artifactcache/casserver.py

@@ -422,7 +422,7 @@ def _clean_up_cache(cas, object_size):
     # which mounts the repo
     stats = os.statvfs(cas.casdir)
     buffer_ = int(2e9)                # Add a 2 GB buffer
-    free_disk_space = (stats.f_bfree * stats.f_bsize) - buffer_
+    free_disk_space = (stats.f_bavail * stats.f_bsize) - buffer_
     total_disk_space = (stats.f_blocks * stats.f_bsize) - buffer_
 
     if object_size > total_disk_space: