Allow configuration of text similarity threshold and improve default

[zmbc]

Addresses #785.

Note that I have conservatively made this change only affect the comparison of cell sources. I think you could make an argument that an analogous change for outputs would also be good, but that would have had much larger impacts on the test suite.

[Carreau]

generally +1 on this, maybe some small refactor and make sure we don't change API.

[zmbc]

@Carreau I believe I've addressed all your comments!

[zmbc]

@krassowski @Carreau what is the next step to move this forward?

[Carreau]

Thanks for your patience, unfortunately the blocking part if maintainer time; I've reaproved the workflow; and thanks for the ping. It might take me some time to get back to it as my actual day-time job takes priority and I have few hours to consecrate to open-source outside of that.

[zmbc]

@Carreau totally understand. Are you open to onboarding new maintainers?

It looks like this PR isn't working on Python 3.8 which is 1.5 years past EOL. Are you open to a PR to remove 3.8 (and 3.9) support?

[Carreau]

Personally yes, but this would have to pass other maintainers as well. But I think it make sens.

[Carreau]

limiting to 3.6 was done as of 83a768f; which was 3 years ago so I think moving to 3.10+ is reasonable.

[Carreau]

and #802 has some discussion where @krassowski is ok droping up to 3.8 (included); so 3.9 is on the edge.

[zmbc]

Ah, thanks for that context. It looks like 3.8 was still included on Windows in error -- I've removed that now.

[zmbc]

The other failing workflow was the "check release" one, with this error:

 M package-lock.json

Traceback (most recent call last):
  File "/home/runner/work/nbdime/nbdime/.jupyter_releaser_checkout/scripts/bump_version.py", line 110, in <module>
    bump(args.force, args.spec)
  File "/home/runner/work/nbdime/nbdime/.jupyter_releaser_checkout/scripts/bump_version.py", line 68, in bump
    raise Exception("Must be in a clean git state with no untracked files")
Exception: Must be in a clean git state with no untracked files

I'm having a hard time understanding what is going on here, and I'm wondering if it has something to do with the fact that this PR is from a fork.