docs(axis): document logMapping and logLinearWidth for log axes

[netscout-mthorn]

Summary

• Add logMapping(string) = 'none' option docs to en/option/component/axis-common.md:
  documents the three values ('none', 'asinh', 'symlog'), tick placement
  behaviour, the splitNumber interaction, and the alignTicks limitation.
• Add logLinearWidth(number) = 1 option docs: explains the quasi-linear region
  half-width shared by both 'asinh' and 'symlog' modes and its effect on the
  linear-to-log transition.
• Update logBase description to clarify that when logMapping is 'asinh' or
  'symlog', logBase controls tick spacing rather than the transform itself.
• Update splitNumber description to note that for logMapping axes, it controls
  how many power-of-base tick candidates are shown, with automatic skipping of
  intermediate powers when the range would produce too many.

All changes are in the shared axis partial (axis-common.md), so they apply to
xAxis, yAxis, angleAxis, radiusAxis, singleAxis, and parallelAxis without
duplication.

Not included

• Chinese translation (zh/option/component/axis-common.md): I attempted
  node tool/patchLanguage --from=en --to=zh but it fails with
  Cannot find module '../editor/common/blockHelper'. This appears to be an
  internal dependency not available in the public repo. I am not able to provide
  a Chinese translation. If a committer is able to apply it, or can advise on the
  expected process for external contributors, that would be appreciated.

Related

• Implementation PR: feat(scale): extend log axis to support negative and zero values echarts#21646

Checks

• npm run build: passes

Questions for reviewers

• The version badges on logMapping and logLinearWidth are currently set to
  6.2.0. Is that the correct target version for this feature, or should they
  be updated?

• The doc text references d3 and matplotlib in two places:

  1. symlog description: "This is the symlog transform used by matplotlib and d3."
  2. logLinearWidth description: "The default of 1 matches the conventions used by d3's scaleAsymlog and matplotlib's AsinhScale."

  The only existing d3 references in this repo (geo-common.md) are functional: they tell users they can use d3-geo directly with the projection API. The references above are citation-style, which has no clear precedent here. Is it appropriate to keep them? A possible middle ground: keep the symlog sentence (it names the term's origin, which aids recognition) and drop the logLinearWidth citation in favour of simply "The default is 1."