Rework extension loading

[baronkobama]

Summary

This PR allows for folders to be loaded (allows for recursing through folders as well) in load_extension and adds the helper method load_extensions that allows for loading multiple extensions in one function call.

Closes #946

Checklist

• If code changes were made then they have been tested.
  • I have updated the documentation to reflect the changes.
• If type: ignore comments were used, a comment is also left explaining why
• This PR fixes an issue.
• This PR adds something new (e.g. new method or parameters).
• This PR is a breaking change (e.g. methods or parameters removed/renamed)
• This PR is not a code change (e.g. documentation, README, typehinting, examples, ...)

Notes

• I've tested this PR myself and haven't found any issues but any additional testing would be much appreciated.
• Credit to @Sodynoizz for a decent portion of the code written here. Assuming this PR looks good, then it will replace Add load_extension_from_path function #1413.

[Sodynoizz]

Should we change load_extensions, load_extension , reload_extension and unload_extension to asynchronous function?

[baronkobama]

PR isn't ready for merge just yet, recursive loading doesn't appear to work currently.

[baronkobama]

Ready for review again, hopefully, bug-free this time around.

[plun1331]

Looks good, I would suggest adding a parameter to ignore (or just print) extension loading errors if multiple extensions are being loaded, or some other solution that doesn't raise an exception and stop extensions from being loaded.

Additionally, we could consider potentially unloading loaded extensions if one fails.

[baronkobama]

Looks good, I would suggest adding a parameter to ignore (or just print) extension loading errors if multiple extensions are being loaded, or some other solution that doesn't raise an exception and stop extensions from being loaded.

The first implementation of this idea is done in b587b10, please take a look when you get a chance. No parameter to change this behavior is added currently but I can add it later down the line if still recommended.

[baronkobama]

No parameter to change this behavior is added currently but I can add it later down the line if still recommended.

This is now implemented in 9e377e7. So far, I've extensively tested the new loading and haven't found any issues. However, due to the addition of numerous new features including a new helper function, recursive loading, and a new parameter, store, any additional testing would be much appreciated.

[baronkobama]

Currently mulling this over and I'm wondering about one last thing before this PR should be fully ready for review/merge: should load_extension be returning the isolated cog name like it currently is or should we be returning the full name that was passed? I'm not sure of this myself but I'm leaning towards returning the full name. Any feedback on this would be much appreciated.

[Lulalaby]

I would prefer full name too tbh

@Dorukyum what you think

[Dorukyum]

I would prefer full name too tbh

@Dorukyum what you think

I agree, it should return the full name.

[Dorukyum]

Looks good to me

[vproyaev]

On line 856 the check is not correct, you check final_out -> it's dict in isinstance(final_out, Exception) it's wrong. Need change to isinstance(final_out[name], Exception)