feat: Add Spanish + Chilean Spanish wordlist

[satelerd]

Summary

Adds ~200 Spanish-language profanity entries grouped into 27 families covering general Spanish, Mexican, Argentine, and (heavily) Chilean slang. Right now devrage shows 0 swears for any Spanish-speaking dev — and trust me, we swear at our coding agents plenty.

What's added

General Spanish:
mierda, puta, joder, coño, cabrón, pendejo, chingar (MX), gilipollas (ES), maricón, verga, polla, boludo / pelotudo (AR), imbécil, idiota, estúpido, tarado, tonto + chat abbreviations (hdp, lpm, ptm).

Chilean (es-CL) — the heavy hitters:

• weón / hueón / huevón family (the Chilean staple)
• aweonao / ahueonado family
• wea / hueá / huevada family
• culiao / culiado family
• conchetumare / conchesumadre + ctm / csm
• chucha, sacowea, cresta, gil, cagada / cagaste, leso

Design notes

• Accent + plain forms both listed — people type weon and weón interchangeably (autocorrect, accent-less keyboards, etc.). Listing both avoids relying on Unicode normalization in the regex.
• Severity reflects usage as a swear, not dictionary meaning. weón is often a friendly filler in Chile (hola weón!) but it's still profanity — marked moderate. Hardcore insults (culiao, ctm, conchetumare) marked strong.
• Group names kept ASCII (weon, cono, conchetumare) so the existing rollup output stays stable across terminals.
• Followed the existing file structure — same comment style, same family groupings, additions placed after the English block with a clear section header.

Verification

• ✅ npm run typecheck passes
• ✅ npm run build succeeds
• ✅ Tested against a battery of Spanish/Chilean phrases — all detected with correct group + severity
• ✅ No false positives on common English words: went (no wea), gilbert (no gil), hello world, plain prose
• ✅ Existing English detection unaffected (fuck this shit still matches as before)
• ✅ Repeated-char + uppercase normalization works (hueeeón, WEÓN MIERDA)

Test transcript

"weón, esto es una mierda"           → 2 matches: weón(weon/moderate), mierda(mierda/strong)
"Ese culiao está bien aweonao, ctm"  → 3 matches: culiao(culiao/strong), aweonao(aweonao/strong), ctm(conchetumare/strong)
"qué wea, conchetumare"              → 2 matches: wea(wea/moderate), conchetumare(conchetumare/strong)
"sacowea hueón, qué chucha"          → 3 matches: sacowea(sacowea/strong), hueón(weon/moderate), chucha(chucha/moderate)
"pendejo cabrón, joder"              → 3 matches: pendejo(pendejo/strong), cabrón(cabron/strong), joder(joder/strong)
"fuck this shit"                     → 2 matches: fuck(fuck/strong), shit(shit/strong)  [unchanged]
"I went to the store"                → 0 matches  [no false positives]
"gilbert had a great time"           → 0 matches  [no false positives]

Happy to split this into multiple PRs (general Spanish vs Chilean, or per-family) if that's easier to review. Cheers from Santiago 🇨🇱

[satelerd]

Self-reviewed against a real 12,508-message corpus (Claude Code + Codex sessions). Found and fixed two regex false-positive classes caused by Spanish accented vowels acting as non-word chars in JS \b:

• cono (de-accented coño) was matching inside ícono / íconos — common in UI/design messages.
• gil / giles / gila were matching inside ágil / frágil / frágiles — endemic in any tech corpus that mentions agile methodology.

Both fixed in the second commit by dropping the de-accented cono (keeping only coño) and dropping the whole gil family with a comment explaining why. After the fix the detector found 1,317 Spanish swears in the corpus that the English-only build missed entirely (it had 121).

Top words discovered (real usage, not meta):

• wea 456 · weon 271 (incl. wn 149, weón 89) · mierda 140 · cagada 140 · puta 80 · chucha 51 · conchetumare 31

Marking as ready for review. Happy to split or trim further if you'd prefer a smaller surface area.