Experiment with linting / applying patches to generated code

[ryan-williams]

Lint fixes, mostly automated using git sub and pre-commit run -a -v:

• Remove many spurious # noqa: … comments
• Remove import six / Python 2 logic
• Remove redundant explicit object-superclass from class definitions

I understand many of these files are auto-generated, but we already apply one patch (post-generate_api).

A possible improvement is to maintain a branch, whose commits represent "patches" we want to apply, and use git rebase instead of apply_json_safe_patch, e.g.:

# Existing commands: regenerate API
download_generator
generate_api "${ABSPATH%/}/openapi-v1.yaml" rest_api
generate_api "${ABSPATH%/}/openapi-v2.yaml" _common.api_v2
run_format

# Commit current changes, store branch name
git commit -am '`update_generated.sh`'
branch="$(symbolic-ref -q --short HEAD)"

# Add commits from "patches" branch
git checkout --detach origin/rw/lint  # any branch will do
git rebase --onto HEAD@{1} `git merge-base HEAD@{1} HEAD`
tree="$(git log -1 --format=%T)"

# "Squash" all patches into initial commit above
git checkout "$branch"
git reset --hard "$(git log -1 --format=%B | git commit-tree $tree -p HEA^ -F-)"

This would allow an arbitrary number of patches to be applied on top of the auto-gen'd docs. Periodically rebasing the "patches" branch (onto main) would serve to resolve rebase conflicts once, allowing subsequent rebases to benefit from them.