Parser refactor (part 1)

[Happypig375]

Now every command is put into a central place for lookup.

Also added \atopwithdelims to let MathListFromLaTeX be able to consume MathListToLaTeX output, and % for comments are now supported in MathListFromLaTeX and MathListToLaTeX.

AliasDictionary has been split into BiDictionary and ProxyAdder.

Contributes to #58.

[charlesroddie]

I had a brief look at this.

• The changes to the data (e.g. adding "\") look fine
• The PR adds net 854 lines of code, and it doesn't reduce complexity. So it starts at -854 points and will need to have very large benefits to justify that. It will take also a long time to review.
• Storing functions in a dictionary, using reference equality, doesn't seem like a great idea to me.
• Why is there Dictionary.cs ? What is wrong with a .Net dictionary?
• Is a Trie needed? Presumably this is for performance? It adds complexity so we would need performance measurements to indicate a parsing bottleneck to justify inclusion. My guess is that parsing latex is not an expensive operation because it can be done from left to right and that rendering is much more expensive. (If string lookup turned out to be expensive, then just separating based on the first character would capture almost all the benefit of a full Trie.)

[Happypig375]

Should include the above justification in a SuppressMessageAttribute.

[Happypig375]

Re-requesting @FoggyFinder's approval since there are a lot of new changes