refactor: split ssr.py into package, enrich OG Online detail pages, fix travel upsert

- Split src/adapters/ssr.py (2160 LOC) into ssr/ package grouped by CMS:
  realworks.py, sure.py, schiedam.py, denhaag.py, overige.py
- Add _og_detail() to api.py; all OG Online scrapers now fall back to
  detail page fetch when energielabel/bouwjaar are missing from the API
- Fix run() to recalculate travel times for existing listings where
  fiets_mark IS NULL; upsert() now writes travel cols on existing rows too
- Update tests/cache.py to patch fetch_soup in every ssr submodule
- Update docs to reflect new package structure and mark API enrichment TODO done

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

CLAUDE.md

@@ -0,0 +1,45 @@
+# Huizenbot
+
+## Doel
+
+Periodiek scrapen van makelaarswebsites in Delft en Schiedam, nieuwe woningen opslaan in SQLite, en pushnotificaties sturen via Home Assistant. Draait als één Docker container op homelab met cron.
+
+Dit draait op het moment al, dus we zijn nu enkel bezig met uitbreidingen en verbeteringen.
+
+
+# AIDE - An IDE for your Agent
+This project uses AIDA to support agents, it increases the robustness of edits and reduces token costs. 
+
+You can always use `aide help` for further info, and use it also on each subcommand. If you do, edit what you learned into the agents.md so we don't have to spend tokens on it repeatedly.
+
+
+## Using aide effectively
+
+**Always start with aide for codebase exploration — not Read or Grep:**
+- Use `aide outline <file>` first to get the function map of any file before reading it
+- Use `aide source <file> <sym>` to read individual functions — never Read a whole large file just to find one function
+- This is especially important for large files like `ssr.py` (84KB+) where Read truncates
+
+**For edits:** `aide insert` is fragile with large inputs (see note above) — fall back to the `Edit` tool for anything non-trivial. `aide replace` is fine for small targeted changes.
+
+## What aide can do (quick reference)
+
+| Command | What it replaces |
+|---------|-----------------|
+| `aide outline <file\|dir>` | `Read` whole file for structure; `ls` + loop |
+| `aide source <file> <sym>` | `Read` whole file for one function |
+| `aide callers <sym>` | `Grep` for call sites |
+| `aide search <term>` | `Grep` across the project |
+| `aide replace <file> <sym> <msg>` | `Edit` / `sed` for symbol-level changes |
+| `aide replace … --lines N-M <msg>` | `Edit` for intra-function line edits |
+| `aide remove <file> <sym>` | Manual splice to delete a symbol |
+| `aide insert <file> <msg> --after <sym>` | Manual splice to add a new symbol — **insert one function at a time**; large messages cause bash to be killed |
+| `aide rename <file> <old> <new>` | Manual find-and-replace of a name |
+| `aide log` | Log related to the undo command; see which files changed in which order |  
+| `aide annotate <file> <sym>` | Persist a non-obvious invariant or gotcha for a symbol |
+| `aide context <file> <sym>` | Read the stored annotation before editing |
+| `aide review [path]` | Check for annotations invalidated by recent edits |
+
+Line numbers in `--lines N-M` are **1-based and relative to the symbol's first
+line** (line 1 is the signature / opening line of the symbol). This means they
+are stable across edits elsewhere in the file.