Since replacing Wikit's Lookup component implementation of SpellingVariantInput in new-lexeme-special-page with Codex's (b8a180d915b80f0ba089cdef465e7bc21ffd4565), the input swallows some characters (see draft changes to cypress tests).
It seems that the first character(s) typed, and sometimes characters typed quickly, confuse the input. Also highlighting the text in the input and trying to write over it reproduces the issue.
Acceptance Criteria
- The codex SpellingVariantInput implementation functions well and does not swallow inputs.