Fix skipping Jupyter cells with unknown %% magic

[AleksMat]

Description

This CL fixes an inconsistency of when to skip formatting of Jupyter cells with unknown magic methods.

Black correctly skips formatting in this case:

%%unknown_custom_magic
<code in another language that black shouldn't format>

However if the cell started with some empty lines black still tried to format it and failed:

%%unknown_custom_magic
<code in another language that black shouldn't format>

Also if the cell started with a comment black tried to format it:

# Some comment
%%unknown_custom_magic
<code in another language that black shouldn't format>

(This example isn't possible in normal Jupyter notebooks but it works in Colab.)

This PR ensures that black will skip formatting cells with a custom magic function which is preceded by any number of empty lines and lines with comments.

Additional notes:

• I moved validate_cell function from __init__.py to handle_ipynb_magics.py module because it belongs there. (It also makes my maintenance of Pyink fork a bit easier if there are less things in __init__.py module.) But if there is a reason to keep it in __init__.py I can move it back.

• Function _get_code_start could also be implemented like this:

  def _get_code_start(src: str) -> str:
      match = re.search(r"^[\s\n]*([^#\s\n].*)$", src, flags=re.M)
      return match.group(1) if match else ""

  It might be a bit more efficient but it is also less readable. Let me know if you would prefer this implementation.

Checklist - did you ...

• Add an entry in CHANGES.md if necessary?
• Add / update tests if necessary?
• Add new / update outdated documentation?

[JelleZijlstra]

Suggested change

	for match in re.finditer(".+", src):
	for match in src.splitlines():

Any reason not to use this?

[AleksMat]

There might be a slight performance difference because re.finditer will stop splitting string once the first line with code is found while src.splitlines() will always split the entire string by lines. But I haven't benchmarked anything and this probably won't affect much the performance of the entire formatting process.

I'm also ok with switching to src.splitlines() if that is the preferred way.

[JelleZijlstra]

Thanks, I guess this is fine.