Debugging linter rules

[awnawab]

This PR adds the following two linter rules:

• ArgSizeMismatchRule: check for size discrepancies between arguments and dummy arguments
• DynamicUboundCheckRule: check for dynamic UBOUND checks for assumed-shape dummy arguments

Other changes include:

• Extension of GenericRule.check and GenericRule.check_subroutine API to allow optional passing of scheduler targets
• Move some of the boilerplate in lint_rules/tests to conftest.py

[github-actions]

Documentation for this branch can be viewed at https://sites.ecmwf.int/docs/loki/116/index.html

[codecov]

Codecov Report

Merging #116 (cb73473) into main (45e51fd) will increase coverage by 0.03%.
The diff coverage is 95.42%.

@@            Coverage Diff             @@
##             main     #116      +/-   ##
==========================================
+ Coverage   92.12%   92.15%   +0.03%     
==========================================
  Files          89       90       +1     
  Lines       16494    16650     +156     
==========================================
+ Hits        15195    15344     +149     
- Misses       1299     1306       +7     

Flag	Coverage Δ
lint_rules	95.96% <95.00%> (-0.71%)	⬇️
loki	92.09% <100.00%> (+<0.01%)	⬆️
transformations	91.78% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
lint_rules/lint_rules/debug_rules.py	94.70% <94.70%> (ø)
lint_rules/lint_rules/ifs_coding_standards_2011.py	97.35% <100.00%> (ø)
loki/lint/linter.py	96.20% <100.00%> (ø)
loki/lint/rules.py	98.57% <100.00%> (+0.10%)	⬆️

... and 1 file with indirect coverage changes

[reuterbal]

This is fantastic! Thanks so much and a perfect start to what I imagine could grow into a quite handy collection of similar utilities.

My main request would be to document this in the loki-lint documentation: https://github.com/ecmwf-ifs/loki/blob/main/docs/source/loki_lint.rst

Otherwise a few comments and questions but nothing that would stop this from going in.

[reuterbal]

[no action required] Just a thought on design: We should probably expose somehow whether a linter rule is changing the IR (and, optionally, call these rules with a cloned IR to ensure rules applied subsequently in the same linter pass get to see the original IR.

[awnawab]

Agreed, great idea!

[reuterbal]

Thanks, very happy with this now.

[reuterbal]

You could even go further and use for carg, rarg in arg_map.items() to save the arg_map[arg] lookups in the loop body. But mentioning this only for educational purposes.