Patch eval metrics to markdown

[EricLiclair]

• adds changes to return metrics from evals
• prints converted metrics to markdown

TODO's:
Collect metrics over the whole benchmark:

• • online,
• • knn,
• • linear, and
• • finetune eval
• Print metrics at end of benchmark script (currently only printing the converted markdown)
• Optional: Print also as markdown table

---

prints the results like:
(values randomly generated and passed to the function)

| Eval Name | Metric Name | Value |
|:---------:|:-----------:|:-----:|
| knn       | val_top1    | 0.44  |
| knn       | val_top5    | 0.27  |
| linear    | val_top1    | 0.26  |
| linear    | val_top5    | 0.27  |
| finetune  | val_top1    | 0.68  |
| finetune  | val_top5    | 0.45  |

which prints markdown like,

Eval Name	Metric Name	Value
knn	val_top1	0.44
knn	val_top5	0.27
linear	val_top1	0.26
linear	val_top5	0.27
finetune	val_top1	0.68
finetune	val_top5	0.45

---

potentially fixes #1333

[guarin]

Thanks a lot for your PR! I left some comments on how I believe we could simplify it a bit but the code overall looks very good.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.90%. Comparing base (9578268) to head (599942d).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1706   +/-   ##
=======================================
  Coverage   84.90%   84.90%           
=======================================
  Files         156      156           
  Lines        6526     6526           
=======================================
  Hits         5541     5541           
  Misses        985      985           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[guarin]

Thanks so much, this looks great!