Make long-running tools fire-and-return to fix intermittent timeouts

redeploy_project, create_project and pull_image used to block the tool
call while polling DSM until the project reached RUNNING / the image
appeared - up to 300s and 240s respectively. On large images this
regularly ran past Claude Desktop's ~4 min tool-call ceiling (the live
MCP log shows redeploy_project calls up to 260s), which the client
reported as a timeout even though the server kept working. Whether a
call crossed the ceiling depended on image size and NAS load, which is
why the failure was intermittent.

These tools now fire-and-return: they trigger the operation, consume
build_stream only for a short early-error window (20s, to catch fast
daemon errors like "manifest unknown"), then return a "running in the
background - check get_project_status / check_image_updates" hint.
Completion is observed via the existing fast status tools.

- DsmClient.trigger_build_stream gains a `budget` parameter
- remove _wait_for_project_running and the _POLL_*/_BUILD_POLL_TIMEOUT
  constants (projects.py) and _PULL_POLL_* constants (registry.py)
- update tests, CLAUDE.md DSM quirks and CHANGELOG; bump 0.7.0 -> 0.8.0

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

CHANGELOG.md

@@ -2,6 +2,38 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.8.0] - 2026-06-16
+
+### Changed (fixes intermittent "MCP not responding" timeouts)
+
+The long-running tools no longer block the tool call until the operation
+finishes. They now **fire-and-return**: trigger the work, surface any fast
+failure, then return immediately so Claude can poll status with the existing
+fast tools. This removes the root cause of the intermittent timeouts —
+analysis of the live MCP log showed `redeploy_project` calls running up to
+**260 s**, which crosses Claude Desktop's ~4 min tool-call ceiling. Whether a
+call crossed it depended on image size and NAS load, which is exactly why the
+failure was intermittent ("works fine for a while, then suddenly times out").
+
+- **`redeploy_project` / `create_project`** — `build_stream` is now consumed
+  only for a short early-error window (`_BUILD_EARLY_BUDGET`, 20 s) to catch
+  fast daemon errors (e.g. `manifest unknown`). If the build is still running
+  after that window (large image), the tool returns a "redeploy/create
+  started — check `get_project_status`" hint instead of polling for up to
+  300 s. The pull and container start continue on the NAS.
+- **`pull_image`** — calls `pull_start` and returns immediately with a
+  "pull started in the background — verify with `check_image_updates` /
+  `list_images`" hint, instead of polling `Image/list` for up to 240 s.
+- **`DsmClient.trigger_build_stream`** — gained a `budget` parameter so
+  callers can cap how long the streamed log is consumed.
+
+### Removed
+
+- `_wait_for_project_running` and the `_POLL_INTERVAL` / `_POLL_TIMEOUT` /
+  `_BUILD_POLL_TIMEOUT` constants in `projects.py`, and the
+  `_PULL_POLL_TIMEOUT` / `_PULL_POLL_INTERVALS` constants in `registry.py` —
+  the minutes-long polling loops they drove are gone.
+
 ## [0.7.0] - 2026-05-18
 
 ### Added