Excluding mountpoint for generating gfid

For generating global file identifier (gfid), unifyfs uses a full pathname including the mountpoint. This work fine when the file system is mounted under a consistent mountpoint (e.g., /unifyfs) across sessions. However, if a user uses different mountpoints across sessions, this becomes a problem:

[root@ninja70 testbed]$ mpirun -hostfile nodelist.txt -n 4 prefix/libexec/sysio-write-gotcha -m /unifyfs
[root@ninja70 testbed]$ mpirun -hostfile nodelist.txt -n 4 prefix/libexec/sysio-read-gotcha -m /unifyfs

Number of processes:      4
Each process wrote:       128.000000 MB
Total reads:              512.000000 MB
I/O pattern:              N to 1
I/O request size:         65536 B
Aggregate read bandwidth: 668.811403 MB/s
Min. read bandwidth:      216.077518 MB/s
Total Read time:          2.369520 sec.

[root@ninja70 testbed]$ mpirun -hostfile nodelist.txt -n 4 prefix/libexec/sysio-read-gotcha -m /unifyfs2
[3] open failed (errno=2, No such file or directory)
[0] open failed (errno=2, No such file or directory)
[2] open failed (errno=2, No such file or directory)
[1] open failed (errno=2, No such file or directory)

When generating gfid (and other places where unifyfs uses full pathnames), the mountpoint prefix should be excluded, i.e., /testfile-0 instead of /unifyfs/testfile-0.

Start by reproducing the behavior with sysio-write-gotcha and sysio-read-gotcha using different mountpoints, such as /unifyfs and /unifyfs2. Trace the gfid generation and other full-pathname handling involved in these commands. Done means the same file remains accessible across sessions when only the mountpoint changes, with the mountpoint excluded from the identifier.