Enforcing PEP8 naming conventions in Parcels

[VeckoTheGecko]

I think we should discuss whether to enforce some of PEP8 variable naming conventions in our codebase.

This can be done via the 'pep8-naming (N)' ruleset from Ruff.

Looking at our current codebase, enabling the N rule with have the following violations

{'N802': 193,
 'N803': 21,
 'N805': 4,
 'N806': 478,
 'N811': 2,
 'N816': 42,
 'N818': 2}

Repro instructions

diff --git a/pyproject.toml b/pyproject.toml
index 86e1420b..4ce41eb6 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -85,6 +85,7 @@ select = [
   "ISC001", # single-line-implicit-string-concatenation
   "TID", # flake8-tidy-imports
   "T100", # Checks for the presence of debugger calls and imports
+  "N", # pep8 naming conventions
 ]
 
 ignore = [

then doing

ruff check . | grep -E '^N\d{3}' > pep-log.txt && python process-log.py

where

# process-log.py
from pathlib import Path

FILE = 'pep-log.txt'

with open(FILE,'r') as f:
    errors = f.read().splitlines()

codes = {}
for error in errors:
    code = error.split(" ")[0]
    if code not in codes:
        codes[code] = 0
    codes[code] += 1

from pprint import pprint

pprint(codes)

As with all Ruff rules, we should be selective about which ones to follow. Some conventions simply do not apply in certain domains (e.g., in numpy, X = np.array(...) is commonly used for multidimensional arrays, and so is dX - even though it violates PEP8) .

In Parcels, we have kernels following CamelCase naming. This is also a pep8 violation as functions aren't named so, but is really useful since its clear what's a kernel and I don't think we should change this convention.

I'll put together a PR that selects which of these rules to enforce and which to ignore - striking a balance that works best for Parcels.