spec/host-tools.json: docker and uv carry no version floor — propose starting points

[ptr727]

Summary

Two required tools in spec/host-tools.json carry "minimum": null, each noting only that no floor has been measured: docker and uv. (git's null is a separate, deliberate case — its why already reasons the floor belongs on git-restore-mtime instead, so it's excluded here.)

Proposal

Not a measured or target floor for either — just a working baseline observed on one machine, offered as a starting point for whoever has broader fleet data to firm it up into a measured or target floor per the convention the rest of the file already follows:

Tool	Observed	Notes
docker	Engine 29.6.2 (docker --version, matching the existing probe/pattern); Docker Desktop app 4.85.0	The host-tools.json probe reads the engine version (Docker version (\d+...) from docker --version), not the Desktop app version winget tracks separately — worth keeping in mind if a floor is added here.
uv	0.12.2 (winget currently offers 0.12.4)	Installed via winget's astral-sh.uv.

Host

Windows 11 Pro 10.0.26200, PowerShell 7.6.4, winget (App Installer), unelevated.

[ptr727]

Fixed in 8cb7e5b (#698), for uv only.

Cross-checked the winget versions here against a second, independently set-up host (this repo's own Linux install path, which always tracks upstream latest): uv reported 0.12.4 there against winget's 0.12.2. The lower of the two becomes a real target floor (0.12.2) in the same style jq's floor already uses, anchored to what an independently set-up host already provides rather than to a measured breakage, with real source/remedy entries per platform.

docker is left at null, deliberately. Unlike uv, it has no remedy anchor anywhere in this repo (it isn't managed by either host-setup tool list, so there's no install/upgrade command to cite), and its existing why already argues the daemon version has never been what broke a run, since the four linters run as pinned images rather than local installs. Per this file's own stated philosophy ("a floor nobody can justify becomes a host failure nobody can act on"), a floor there would be unfounded rather than a starting point worth firming up. If broader fleet data ever turns up an actual docker version defect, that's the case for a measured floor, not a version comparison across two working hosts.