Fix multiline semantic token encoding in heredoc

[arnaud-dezandee]

Summary

Alternative to #2095.

This fixes the same multiline semantic token encoding bug, but takes a slightly broader approach. In addition to fixing the bad delta anchor after multiline token emission, it also fixes how multiline token segments are sliced per line.

I hit this while reproducing broken Terraform heredoc highlighting in VS Code.

What changed

• track the last emitted token position directly
• compute multiline segment lengths from the correct per-line start offset
• skip zero-length segments
• add regression coverage for the heredoc interpolation case

Why not just #2095

#2095 fixes the overflow symptom. This PR is an alternative that also fixes the multiline slicing itself. While testing, I found the encoder was over-reporting the first emitted line length for multiline tokens that start late on a line, so I updated the existing multiline test as well.

Validation

• reproduced locally in VS Code
• verified the highlighting fix with a custom terraform-ls binary
• ran focused tests
• ran the full terraform-ls test suite

PCI review checklist

• I have documented a clear reason for, and description of, the change I am making.
• If applicable, I've documented a plan to revert these changes if they require more than reverting the pull request.
• If applicable, I've documented the impact of any changes to security controls.
  Examples of changes to security controls include using new access control methods, adding or removing logging pipelines, etc.