Parser.Range used len(data)-len(highlight), which only matches suffix
slices. Single-byte highlights like b[0:1] are subslices of the
remaining buffer, so the wrong offset pointed at the end of the
document (issue #1047). Use pointer-based subslice offset like
wrapDecodeError.
Add regression tests for unstable.Parser and toml.Unmarshal error
positions and human-readable output.
Co-authored-by: Thomas Pelletier <thomas@pelletier.dev>
Cursor Agent