zstd benchmark mode: better progress indicator #4850

Currently on large corpuses (-r with large of many files) zstd's benchmark mode (-b) can reasonably look like it got stuck:

• first it ingests the corpus, printing the name of each file
• then nothing happens (it kinda looks stuck on the last direntry) until the first compression level completes
• same for second
• ...

Would be nice if the command-line harness provided some signs of life and some progress e.g. [N/K] [ xy%] where K is the number of levels in the benchmark (in -b/-e mode), N is the currenth, and xy% is the progress for the current. -S exists but it's a bit too detailed.