tiny-count: fix handling of gzipped tiny-collapse outputs for SummaryStats

[AlexTate]

The SummaryStats class now properly handles gzipped tiny-collapse outputs when the user elects to have them produced. Previously, it assumed all tiny-collapse outputs to be plaintext

Closes #221

[AlexTate]

Leaving this PR as a draft during final testing.

An issue remains: SummaryStats for standalone runs that source third party inputs. Currently:

• At worst: the user follows the same filename suffix conventions as tiny-collapse and places these outputs in their CWD when executing tiny-count. This will lead to a crash. To help, we could re-implement an old policy of requiring the --is-pipeline flag before even searching for pipeline outputs in the CWD. This would work only as long as the user is honest with the flag...
• At best: the above requirements aren't met and tiny-count assumes a non-pipeline mode, and in this case it doesn't bother parsing tiny-collapse outputs in the CWD