Enable PYINPUT Processing Through Common Python Interpreter

[bska]

This PR allows client code to use a single Python interpreter for all input parsing operations. Having an independent interpreter as part of the internal ParserState class means we run the risk of failing to initialise that interpreter if the calling code has already created a separate interpreter object. That, in turn, means that the ability to parse PYINPUT is generally dependent on the client code's environment so we need a way to inform the parser of that environment.

The default remains that the Parser creates an independent interpreter object.

[bska]

jenkins build this please

[bska]

jenkins build this please

[bska]

jenkins build this please

[bska]

jenkins build this please

[akva2]

LGTM. I do slightly feel like it would be more consistent to just flat out require a pre-created interpreter when support is enabled, but it would of course break existing code.

[bska]

I do slightly feel like it would be more consistent to just flat out require a pre-created interpreter when support is enabled, but it would of course break existing code.

I agree, and I think we could even take it one step further and to have a system-wide singleton for the Python interpreter instead of passing it as an argument into functions/constructors. Given that there must be exactly one interpreter for the entire process I think a singleton makes a certain amount of sense for that.

That's a (much) bigger change, however, so I'll do that as follow-up work if we decide to do it at all. For now I'll merge this since the PR's approved and build check is green.