Fixes #366 Seemingly unnecessary double quoting and escaping

[schuylermartin45]

Fixes a number of issues around excessive quotes in JINJA statements and other support. This should significantly improve previously seen odd behavior in crm convert and crm bump-recipe.

Also adds:

• Parser support for multiline strings that use the "double-quote-and-backslash" method
• Introduces a new NodeVar class to allow us to support tracking comments and selectors on JINJA variables. Most of this support is for V0, but there is some for V1.
• Improves a number of critical JINJA regular experessions
• A number of bug fixes around multiline string parsing
• A minor refactor of some older test files
• Regression tests that validate all these other improvements

From the integration tests, it looks like there is a slight decline in compatibility (by maybe a few recipe files). In my investigation, it looks like the recipe files that are now failing, for the most part, already contain invalid YAML. Overall, this should be a massive improvement over the current parsing issues.

[schuylermartin45]

This is why we have such lovely integration tests:

  "exception_histogram": {
    "EXCEPTION: An exception occurred while parsing the recipe file": 238,
    "EXCEPTION: An exception occurred while converting to the new recipe file": 15
  },

The change in its current state tanks the success rate for parsing AnacondaRecipes and conda-forge recipe files.

[schuylermartin45]

It looks like comments/selectors on JINJA assignment lines are causing parsing issues with the current changes.

{% set native_compiler_subdir = 'linux-64' %}  # [linux]

[schuylermartin45]

A brief test shows that crm convert has apparently always dropped JINJA variables with comments on the same line. That has gone undetected for a while now and is tied to the variable assignment logic.

I won't break this out into a new ticket, this needs to be addressed in these changes.

[skupr-anaconda]

Nice. The PR will fix this extra double-quoting https://github.com/AnacondaRecipes/pyside6-feedstock/pull/2/files#diff-f3725a55bf339595bf865fec73bda8ac99f283b0810c205442021f29c06eea9aR2, right?

[schuylermartin45]

Nice. The PR will fix this extra double-quoting https://github.com/AnacondaRecipes/pyside6-feedstock/pull/2/files#diff-f3725a55bf339595bf865fec73bda8ac99f283b0810c205442021f29c06eea9aR2, right?

It was supposed to, but I just looked and the test regression is wrong. This must have gotten lost in the fray, I'll take another look.

[schuylermartin45]

@skupr-anaconda Now we should be good. At least for the vast majority of cases.