[PDF OCR] Remove the assumption of language matching

Remove the assumption that the language for WordBox should be matched
with that for its enclosing InlineBox. This assumption can't be always
true due to:
(1) PDF can have words from different languages within a sentence.
(2) OCR may misrecognize a word's language.

AX-Relnotes: n/a.
Bug: 1423809
Change-Id: Iea8e10ae56b4fba3260648238aa6fbeca23777b4
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4398790
Reviewed-by: David Tseng <dtseng@chromium.org>
Commit-Queue: Kyungjun Lee <kyungjunlee@google.com>
Cr-Commit-Position: refs/heads/main@{#1126338}