Remove Python2 style super()-calls and fix ExtensionEasyBlock

[Flamefire]

This removes a lot of "clutter" and avoids C&P mistakes when extending or adding new easyblocks based on existing ones

The test suite then failed due to a bug in our usage:

• ExtensionEasyBlock overrides make_module_extra from EasyBlock
• It replaces the altroot/altversion parameters by a single "extra" so when calling it now with the expected arguments, e.g. via super it fails with a TypeError.

Will be fixed and hence requires

• Remove Python2 constructs easybuild-framework#4834

[Flamefire]

@Micket rebased after merge of the framework PR

[boegel]

@Flamefire So with the framework PR merged, we now have a bug in some easyblocks (the ones that derive from ExtensionEasyBlock, which is being fixed in this PR?

[boegel]

@Flamefire So with the framework PR merged, we now have a bug in some easyblocks (the ones that derive from ExtensionEasyBlock, which is being fixed in this PR?

After checking the framework PR (easybuilders/easybuild-framework#4834) again, it seems like the incorrect usage of just passing extra to make_module_extra is still supported, but deprecated.

At first sight it seems this is only relevant to the RPackage easyblock.
@Flamefire Is that correct, or were there other cases?

[Flamefire]

@Flamefire So with the framework PR merged, we now have a bug in some easyblocks (the ones that derive from ExtensionEasyBlock, which is being fixed in this PR?

It is actually the other way round ExtensionEasyBlock is in framework and some easyblocks make use of the running into the issue where the override change parameters. (There is actually a PyLint warning for this as usually an override should act "similar enough" to the base class [LSP])

At first sight it seems this is only relevant to the RPackage easyblock.

I'm not sure if there are others in e.g. custom easyblocks hence the deprecation only similar to other changes we did. I fixed the RPackage easyblock to avoid triggering our own deprecation warning.

[Flamefire]

@lexming Thanks for the probably exhausting review.

I double checked the instances where there would be a change and reverted them or explained why it makes sense. Please see if you agree.

[lexming]

Thanks for the update @Flamefire. I agree with the changes/fixes you propose to those super calls that now jump to higher parents, but those changes complicate merging this PR as we should test them. I'll start running those tests and uploading reports.

[lexming]

Test reports for those easyblock with changes beyond replacing super calls to immediate parent with super():

• ✅ Perl-bundle-CPAN/5.38.2-GCCcore-13.3.0: https://gist.github.com/lexming/fe8a4434f70f80b45c107d915f7aa7d1
• ✅ XML-LibXML/2.0210-GCCcore-13.3.0: https://gist.github.com/lexming/783ca8ea21018574a983fa64b74f0c01
• ✅ Python-bundle-PyPI/2024.06-GCCcore-13.3.0: https://gist.github.com/lexming/c8b05e80464ab5ee169d3cdea1727ee9
• ✅ IPython/8.28.0-GCCcore-13.3.0: https://gist.github.com/lexming/69b1a7c6e48265d7fc94362c22fe64f4
• ✅ imkl-FFTW/2023.1.0-iimpi-2023a: https://gist.github.com/lexming/cce1958504362262e01c36d130aa4fee
• ✅ OpenSSL/1.1: https://gist.github.com/lexming/68f6a243911d640b45657d19080deadc
• ✅ Tkinter/3.11.3-GCCcore-12.3.0: https://gist.github.com/lexming/443d408269ec216916ae825c0e178080

[lexming]

LGTM

[lexming]

Merging, thanks @Flamefire !