docker: production image + Gitea Actions for monthly refresh

Dockerfile: self-contained image with corpus + Chroma + BM25 baked
in. Drawbar's compose pulls + runs without volume mounts. Built from
sources.json (labels schema), PRODUCT_NAME=crop_chem by default,
HYBRID_SEARCH=true (always-on for production quality). RERANK_URL +
OLLAMA_URL get set at compose time.

.gitea/workflows/refresh.yml: monthly cron (1st @ 06:00 UTC) does
full scrape → reindex → image push. Scrapes Bayer (~30 min) +
EPA PPLS row-crop filtered (~7h). Skips reindex+push if no corpus
diff. Tags pushed: :latest, :<sha12>, :corpus-<YYYY.MM.DD>.

.gitea/workflows/image-only.yml: on-demand or auto on code-only
pushes to main (paths: docs_mcp/, rag/, scrape/, requirements.txt,
Dockerfile, sources.json). Reindexes from committed corpus, builds
image, pushes. ~10 min vs ~9h full refresh.

.gitignore: corpus/ now COMMITTED (4,159 labels, 265 MB of .md +
sidecars). Lets image-only.yml rebuild indexes without re-scraping.
chroma/ + bm25/ still gitignored (regenerable binary indexes).

.dockerignore: drops venv, eval results, PLAN/README/CLAUDE.md,
deploy/, .git/ — keeps the image lean. corpus + chroma + bm25
explicitly NOT in dockerignore (those go INTO the image).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

corpus/epa_ppls/241-297.md

@@ -0,0 +1,45 @@
+# SQUADRON HERBICIDE
+
+- EPA Reg No: **241-297**
+- Registrant: BASF AGRICULTURAL SOLUTIONS US LLC
+- Signal word: Danger
+- Active ingredients: Pendimethalin (21.85%); 3-Quinolinecarboxylic acid, 2-(4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl)-, monoammonium salt (3.84%)
+- Label accepted: 1991-09-03
+- Source PDF: https://www3.epa.gov/pesticides/chem_search/ppls/000241-00297-19910903.pdf
+
+---
+EPA Rag. No. 241-297 
+FOR USE ON SOYBEANS 
+SURFACE APPLICATIOr ~a UP TO 30 DAYS PRIOR TO PLANTING 
+OBSERVEALlPRECAUT~YSTATEMENTSINTHEsauADRONLEAFLETlABEl.SEETHESQUADRON 
+LABEL FOR MIXING. SPRAYING. APPLICATION INFORMATION. USE RATES. AND ROTATIONAL CROP RE­
+STRICTIONS. WHEN TANK MIXING. ALWAYS FOLLOW THE MOST RESTRICTiVE lABEL. 
+DfREcnONS FOR USE 
+h is a violation 01 Federal law to use this product In a manner inconsistent with lis labeling. DO NOT apply this procb:t 
+through any type of irrigation system. ThIs 1ab61 must be In the possess'on of the use . at the time of helblclcle 
+application. 
+SQUADRON can be surface applied at planting or up to 30 days prior to pIanIing. Preemelllence appllcalions of 
+SQUADRON or SQUADRON plus PROWl- herbicide as clifec:ted by this label wia control weeds as .sted on the 
+SQUADRON leaflet label. 
+AddItional PROWl.. may be required tor certain soils. For rates refer to the sauADnON plus Additional PROWl 
+herbicide Tank Mixture section of the SQUADRON label. 
+Rainfall (approx.1 Inch) is required to activate surface applied herbicides such as SQUADRON before weed 
+emelQ8llC8. If adequate moisture Is not received within seven clays after application. a thorough shallow tillage or 
+Q,Il1lvatiofl (preferably with a rotary hoe) may IlI1JI'Ove weed control. 
+To kiD existing vegetation SQUADRON may be tank mixed with paraquat or Roundup'. Consuh labels for mIlS 8nI! 
+use InstNdIons. Mairjaln constant agilalion of spray mixlure. . 
+Large seeded weeds IUCh u Eastem black nightshade and giant ragweed germinating deep In the soli or with 
+subsequent flushes may require a Qlllivation or postem&lQ8llC8 herbicide application for season IoQI1 C{lI1Irol. 
+• c. ,,..,,... . .. 
+Use of SQUADRON heIbIcIde In accordance with label directions is expected to result'" ~nhai growIr.of rotational 
+crops in most duations; l1rwever. various environmental and agronomic factors mak~'.~sIIIp<tp Il{n'IIna1e al 
+risk. associated with the use of this procb:t and, II1ereIore. rotatIOnal crop Injury Is alwall ptlsslli6. • c 
+ACCEPTED 
+SFP - 3 \99\ 
+.. 
+." -, , .. ,.". f. " , ( 
+•• • , .. 
+,. < .. , . ,. 
+< , 
+, • 4 • , , 
+.. -___ . 'P---"-~--"-\"~-~'- ~ ..... _ ... " . ..,... ... ~ - • ~. .-. 1 ___