New journey search string parser

- The old code was redundant and not easy to read
- There are now three analyzer steps: Lexically, syntactically and
contextually (checking keyword order and other input errors).
- One class JourneySearchAnalyzer, that uses these three analyzers to
parse an input string and update the cursor position if there are
new items inserted (correction) before the current cursor position
- Not ready yet: Completion for the "at" keyword works, for "in" it has
to be implemented. Some other mistakes like double quotation marks
after clicking on a stop suggestion. Another TODO are more/better
corrections.
- The new parser should be quite performant (optimized with help of KCachegrind)