@laco0416 is working on making the interpolation config configurable in #9158.

Once this PR lands we can work on further Lexer/Parser improvements.

We should add a new INTERPOLATION_TOKEN to the lexer output for interpolations.

Benefits:

• Any character is allowed in an interpolation - currently < in an interpolation is considered as a tag start and we backtrack on malformed track. a <b &c> d would defeat our current strategy. (addresses in Improve lexers/parsers #9421)
• There is currently no way to escape ICU messages (expansion forms) and interpolation. Because the lexer would be responsible for interpolation and expansion form, we can implement escaping by using %xx for those.