Use byte offsets for internal ranges

[msujew]

To reduce on memory usage, this change removes the line/column information for all internal text ranges and only leaves the byte-offset ranges in place. On one hand, this simplifies some code, while making some other more complicated (especially LSP response handling).

Reduces the memory usage of tokens by 20% (16 bytes). Also removes 16 bytes from all AST nodes. Since we no longer have to track utf16 line/column infos, this speeds up the lexer by roughly 10%.

[sailingKieler]

The change looks pretty good to me, and makes sense.
In practice, I assume only the workspace symbol and document symbol providers might notably be affected performance-wise. In critical cases we always can modify the concrete ...Impl struct generation to mitigate potential runtime performance issues - if such ever show up, right?

I have just some API-related hints/questions.

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.50.

Benchmark suite	Current: b69a4a6	Previous: d803417	Ratio
BenchmarkWorkspaceCycle (typefox.dev/fastbelt/examples/statemachine) - MB/s	12.98 MB/s	5.21 MB/s	2.49

This comment was automatically generated by workflow using github-action-benchmark.

[sailingKieler]

This looks really nice, many source portions feel much more streamlined.

I have just two very minor points, see below. 🚀