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/352-470.md

@@ -0,0 +1,78 @@
+# DUPONT BLADEX 4L HERBICIDE
+
+- EPA Reg No: **352-470**
+- Registrant: CORTEVA AGRISCIENCE, LLC
+- Signal word: Warning
+- Active ingredients: Cyanazine (43%)
+- Label accepted: 1999-01-15
+- Source PDF: https://www3.epa.gov/pesticides/chem_search/ppls/000352-00470-19990115.pdf
+
+---
+PM 2.6" ~'S ;2.-410 SUPPLEMENTAL LABELING n \ & ' 
+l/15/~~ 
+DuPont Agricultural 
+Products 
+.... A Growing Partnership With Nalure" 
+BLADEX® 4L HERBICIDE 
+MAXIMUM USE RATES FOR 
+1999 - 2002 
+BLADEX® 4L HERBICIDE 
+EPA REG. NO. 352-470 
+MAXIMUM USE RATES FOR 1999 - 2002 
+DIRECTIONS FOR USE 
+f is a violation of Federal law to use this product in a manner 
+mconsistent with its labeling. 
+DuPont BLADEX 4L Herbicide is a liquid formulation that can 
+be applied alone or in combination with other herbicides to 
+control a wide variety of weeds in field corn, sweet corn, 
+popcorn, field corn grown for seed, and cotton. The use rates 
+specified in this supplemental label supercede all specimen and 
+packaging labels for the year 1999. 
+CYANAZINE RATE LIMITs 
+One quart of BLADEX 4L contains 1.0 Ib cyanazine active 
+ingredient (aj.). Rate limits for 1999 - 2002 described below 
+apply to cyanazine from all sources. Adhere to the rate recom­
+mendations in this label and observe the rate limits and other 
+,f"strictions for the time period indicated. 
+1~ 
+~ 
+1999: 
+• Do not apply more than 3.0 quarts of BLADE X 4L per acre per 
+year. 
+• An enclosed cab is required when applying BLADEX 4L. An 
+enclosed cab must have a nonporous barrier that totally sur­
+rounds the occupants and prevents contact with pesticides 
+outside of the cab. 
+2000-2002 
+• Do not apply more than 1.0 quart BLADEX 4L per acre per 
+• This product may not be sold or distributed after September 30, 
+2002. 
+• This product may not be used after December 31, 2002. 
+When state/local requirements regarding the use of cyanazine 
+(inCluding lower maximum use rates andlor higher set-backs) 
+differ from the label, the more restrictive/protective require­
+ments apply. 
+IMPORTANT PRECAUTIONS 
+Refer to the BLADEX 4L label for weed control, use rate, 
+mixing and other use instructions. Use rates vary according to 
+application type, timing and soil type. Read and follow all use 
+directions and precautions but do not exceed the maximum rates 
+listed above. 
+IMPORTANT 
+BEFORE USING "BLADEX 4L" READ AND 
+FOLLOW ALL APPLICABLE DIRECTIONS, 
+RESTRICTIONS AND PRECAUTIONS ON THE 
+EPA-REGISTERED LABEL. 
+This bulletin contains new or supplemental instructions for use 
+of this product which does not appear on the EPA-registered 
+package labels. Follow these instructions carefully. 
+This label must be in the possession of the user at the time of 
+application. 
+BLADEX 4L is a Restricted ~ J!l PTE D 
+year to any crop land. JAN J 5 1!J99 
+D-111198 1Jndor "'" hIoraI _'rb • An encloscd cab is required when applying BLADEX 4L. Pw>g\o\de, &lid 1' __ Ad., 
+.. __ , tor "'" ~ 
+H-
+For product information call1-888-6-DUPONT ~~ () 
+© 1998 E. I. du Pont de Nemours and Company, Agricultural Products, Wi mmg on, e a 
+Page 1 of 1