Letter level word markup with accents #324
Description
The current setup for this is not truly robust.
- It relies on the definition of the word constituents in the splitting algorithm (around lines 880,881) to be compatible with the pattern used for getting the constituent letters (around lines 3721,3722 to be) - if there is a mismatch, letters might get lost.
- currently only accent commands \", \', \` and \~ are recognised, and only, e.g. \"a is interpreted as letter but not \"{a} or "a (the latter from only allowed in germanb babel)