[Docs] Replace the universal distance threshold table

Branch: docs/distance-thresholds onto main | Workspace: dev/audit-fixes

The distance reference claimed 100.0 means "exact opposites", which is not a bound of any Delta E algorithm.

Changes

• docs/reference/distance.md: drop the two unfounded rows, state that the scale is algorithm-dependent, and add the measured figures.

Notes

• Black against white scores exactly 100 in CIEDE2000, which is where the claim came from, but that is the lightness axis, not a maximum: green against magenta reaches 108.6.
• The same pair scores 100 in DeltaE94 and 97.9 in CMC, and their largest measured distances are 148.9 and 207.7.
• The first three rows are kept: they are the usual CIEDE2000 reference points and are not in question.
• Readers are pointed at calibrating thresholds on their own samples.