Interpreter Refactor

[florianessl]

• Refactors & restructures some of the logic of my "AntiLag" branch into a new "Caching" namespace.
• Moves some of the common helper function from the interpreter class to a new file "game_interpreter_shared.cpp"
• Adds a new abstract base class for the interpreter: "Game_BaseInterpreterContext"
  • This class holds the declarations for a few member functions which are used for evaluation logic across different sorts of commands
  • All the moved helper functions (CheckOperator, ValueOrVariable, CommandStringOrVariable & others) are agnostic to the actual implementation of the interpreter itself
• Refactors "ValueOrVariable" & "ValueOrVariableBitfield" into template methods
  • Might save some unecessary branching depending on the context in which they are used

New helper function: "DecodeTargetEvaluationMode"

• Unifies some of the commonly seen logic that was responsible for decoding command arguments into either a single variable id or a set range of variables.
• Used in ControlSwitches, ControlVariables & ControlStrings with different template parameters (these specify the supported variable accessing modes). E.g.: A hypothethical new "ControlVariablesEx" command could just set all of these parameters to true and share the rest of its logic with the original command.
• Used in my ScopedVars branch for these new commands:
  • ControlSwitchesEx
  • ControlVariablesEx
  • ControlScopedSwitches
  • ControlScopedVariables

This should be most of the easier-to-verify-and-merge changes, my "ScopedVars" depends on.

The current draft for ScopedVars also contains some more changes for the "ControlVariables" & "ConditionalBranch" commands, moving all of the operand switching logic into game_interpreter_control_variables.cpp. Basically I reinvented dispatching tables for this, to share this logic between the different new command spin-offs, but I have yet to confirm with benchmarks that this really a good solution. I don't want to slow any of the existing commands down by even a single cycle. :D
https://github.com/florianessl/Player/tree/ScopedVars
Best possible way to implement all this switching would likely be a mix of computed go-tos & tail recursion (see #1919 for the suggestion of implementing computed go-tos; but it´s debatable, if any of these changes would create any speed-ups in current-year & aren't already part of common compiler optimizations.)

[Ghabry]

Nice refactor / cleanup! Besides Refactored several Interpreter related code segments this does not look too controversial. Reviewing that one commit will take a while but on first glance I see no obvious issues.

[Ghabry]

@florianessl could you attempt to do a rebase? There is one conflict in game_interpreter.

[Ghabry]

Would be good to get this one reviewed and in before the event commands because of conflicts.

Will run our test game through this to find bugs...