Fix metadata parsing

[thorbjoernl]

Change Summary

Due to some unfortunate choices of separation characters in file names, it is difficult to properly derive the metadata of a file from the file path alone in some instances. Here I implement an ugly hack to make the parsing work for old templates, while changing the template (and encoding scheme) to prevent this issue in future experiment runs.
The hack relies on some assumptions:

• network must not contain _ but can contain -
• region can contain _

This makes some changes which may break stuff, but I think it is necessary to be able to implement proper querying (see #117) which will allow us to remove a lot of the filename parsing in pyaerocom (which leads to breakage when aerovaldb changes storage location).

For future runs, a new template string will be used which consistently uses _ as separation character in filenames. Should _ be included in an argument it will be encoded.

For context, this goes together with the following PRs, but trying to break it into smaller chunks for easier review:

• Implement new query interface for aerovaldb #121
• Implement areovaldb querying for accessing file metadata in pyaerocom pyaerocom#1529

Related issue number

closes #119

Checklist

• Start with a draft-PR
• The PR title is a good summary of the changes
• PR is set to AeroTools and a tentative milestone
• Documentation reflects the changes where applicable
• Tests for the changes exist where applicable
• Tests pass locally
• Tests pass on CI
• At least 1 reviewer is selected
• Make PR ready to review

[heikoklein]

Looks good.

Will you add the new character-constraints to pyaerocom, too, to get a proper feedback if wrong project/experiment name is chosen?

[thorbjoernl]

Will you add the new character-constraints to pyaerocom, too, to get a proper feedback if wrong project/experiment name is chosen?

I don't think I am introducing any new character constraints, but rather following the unwritten constraints as they currently exist. The pyaerocom splitting of filenames would already break if these constraints aren't followed.

That being said, as part of #121 I support transparently encoding characters that aerovaldb requires for disambiguating metadata in file names which removes these character constraints moving forward. Maybe we can revisit the discussion on that PR?