Add completion possibilities that split the current token

Consider the following (where the vertical line represents the cursor position):

program = dr|codeWorldLogo

Triggering autocomplete will now try to match the entire token, `drcodeWorldLogo`. Autocomplete will decide that `drawingOf` is the best suggestion by far, and turn the whole thing into `drawingOf`. The `codeWorldLogo` is lost.

Instead, auto-complete should be willing to also consider splitting the token into `dr` (before the cursor) and `codeWorldLogo` at the end. If they are BOTH close to existing tokens, then autocomplete should potentially replace the `dr` with `drawingOf`, and leave `codeWorldLogo` alone. The decision about whether to split the current token or not should be made based on the edit distance to turn the portion after the cursor into a known token, as compared to the edit cost of deleting it after this completion, possibly with some weighting in favor of one or the other choice.

Im Issue werden keine Implementierungsdateien oder Tests genannt. Beginne damit, den Autocomplete-Code zu finden, der zum aktuellen Token und seiner Edit-Distance-Logik passt, und verfolge anschließend, wie der Cursor das Token aufteilt. Fertig ist die Arbeit, wenn Autocomplete den Ersatz des gesamten Tokens mit der Aufteilung vor dem Cursor vergleichen, ein gültiges Suffix beibehalten und konsistent zwischen den Alternativen wählen kann.

Vom Indexierungsmodell aus dem Issue-Text verfasst.