refactor(flake): user registry, multi-user hosts, and portable home outputs (#49)

## Summary

Separates user identity (data) from the reusable Nix modules and lets a host declare any number of users, replacing the previous one-user-per-host structure. Also restructures the tree and exposes the home config for use off these hosts.

## Changes

- **User registry** (`users/registry.nix`): per-user identity (name, email, groups, authorized + signing keys) as the single source of truth; no user data hardcoded in modules.
- **Multi-user `mkHost`**: a host declares a `users` set keyed by username; per-user identity is injected into each home config via the `identity` module arg.
- **Restructured layout**: `users/`, `home/`, `modules/`, `hosts/`, `lib/` replace the former `lyrathorpe/` and `system/` trees.
- **Portable outputs**: standalone `homeConfigurations."<user>@<system>"` (the portable subset — shell, git, editor, claude) plus an exported `homeModules` for use on machines not managed by this flake, or as an input to other flakes.
- Docs (`README.md`, `home/README.md`) and `.gitignore` updated for the new paths.

## Fixes

- Closes #46 — shared user module authorized one user's SSH key for every account.
- Closes #47 — git committer identity hardcoded as defaults instead of per-user.
- Closes #48 — EDaaS systemd linger hardcoded to a literal username.

## Verification

- `nix flake check` passes: treefmt, deadnix, statix, pre-commit, and evaluation of all NixOS hosts + Darwin + homeConfigurations.
- Derivation-path comparison vs `main`: `lyrathorpe-mbp` and `emmathorpe-edaas` are byte-identical; `lyrathorpe-t400`, `lyrathorpe-macpro31` and `lyrathorpe-rpi5` differ only by de-duplicating a repeated `authorized_keys` entry (confirmed with nix-diff — no other change).
- Standalone `homeConfigurations."lyrathorpe@x86_64-linux".activationPackage` builds.

## Notes

- `emmathorpe` has no personal authorized key yet (it previously inherited Lyra's key via the bug in #46); the registry entry is intentionally empty — add a real key if SSH login as `emmathorpe` is wanted (moot on the WSL host).
- A two-repo (public dotfiles / private systems) split is deferred by design; this internal restructure is the prerequisite for it.

---------

Co-authored-by: Emma Thorpe <emma.thorpe@citrix.com>
Reviewed-on: #49

hosts/RPi5/docker.nix

@@ -0,0 +1,34 @@
+# Docker host with the daemon socket exposed over the network.
+#
+# SECURITY: the daemon listens on plain TCP 2375 with NO TLS and NO auth. Access
+# to that port is root-equivalent on this host (the Docker API can mount the
+# host filesystem and run privileged containers). The ONLY thing protecting it
+# is the nftables rule below, which accepts 2375 solely from the trusted LAN
+# subnet. Do not widen that subnet to anything you do not fully trust. The
+# secure upgrade path is mutual TLS on 2376 (--tlsverify with client certs);
+# that needs out-of-band cert provisioning and is intentionally not wired here.
+{ ... }:
+{
+  virtualisation.docker.enable = true;
+
+  # Expose the daemon over TCP by extending systemd socket activation rather than
+  # setting daemon.settings.hosts. The NixOS docker unit starts dockerd with
+  # `-H fd://` and takes its listeners from this socket; putting `hosts` in
+  # daemon.json as well would conflict with that and dockerd would refuse to
+  # start. Adding the TCP listener here keeps a single source of truth.
+  # The leading "" resets the unit's default (unix-socket-only) ListenStream list.
+  systemd.sockets.docker.socketConfig.ListenStream = [
+    ""
+    "/run/docker.sock"
+    "0.0.0.0:2375"
+  ];
+
+  # Source-restricted firewall rule for the Docker TCP port. 2375 is deliberately
+  # NOT added to networking.firewall.allowedTCPPorts (that would open it to every
+  # source); instead nftables accepts it only from the trusted subnet. Adjust the
+  # CIDR to match the LAN that should reach the Docker API.
+  networking.nftables.enable = true;
+  networking.firewall.extraInputRules = ''
+    ip saddr 10.187.1.0/24 tcp dport 2375 accept
+  '';
+}