Line-by-line parsing

If we introduced parsing individually for each `lineitem`, we could very easily cache and reuse prefix parses by storing the full compiler state after the lineitem. Would require adjusting `loc`s.

The issue names no files, tests, or entry points, so begin by locating the compiler parsing implementation and its handling of lineitems and locs. Done means parsing lineitems individually, caching and reusing prefix parses through the full compiler state, and adjusting locs without breaking existing behavior.

Written by the indexing model from the issue text.