Spurious gap between an abbreviation period and the sentence-final period (Jr. . instead of Jr..)

When a sentence ends right after an abbreviation (\`Jr.\`, \`etc.\`, \`D.C.\`, ...), Brown's own convention keeps both the abbreviation's period and the sentence-final period as two adjacent, flush characters (\`Jr..\`) rather than eliding one -- confirmed directly against \`src/semcor/brown-nolines.txt\` (e.g. \`Mayor-nominate Ivan Allen Jr..\`). This corpus's own data has 154 raw candidates of an abbreviation-ending token immediately followed by a separate \`.\` token with a single spurious space between them (\`Jr. .\`).

Context-window verification against \`brown-nolines.txt\` (same technique as the #8 quote-gap follow-up and #43's hyphen-dropped-word-pair fix) confirms 100 of these unambiguously. The other 54 are left unresolved for now, mostly because the abbreviation itself is a multi-word underscore-joined token (\`N._Y.\`, \`D._C.\`, \`U._S.\`) where Brown's real text has *no* space at all between the parts either (\`N.Y..\`, not \`N. Y..\`) -- a related but distinct bug in how this corpus's underscore-joining convention handles this class of abbreviation, worth its own separate look rather than folding it into this fix.

Start by reading src/semcor/brown-nolines.txt and compare the 100 unambiguous abbreviation-period candidates against the Brown source context. Trace the repository's corpus correction or generation workflow before editing, and verify that those cases use adjacent periods while the 54 unresolved multi-word abbreviation cases remain out of scope.

Written by the indexing model from the issue text.