Task #252 — docs/ index + parent doc for the driver CLI suite

Per-CLI runbooks (Driver.{Modbus,AbCip,AbLegacy,S7,TwinCAT}.Cli.md) shipped
with #249-#251 but docs/README.md's Client tooling table never grew entries
for them and there was no parent doc pulling the suite together.

Adds:
  - docs/DriverClis.md — short parent. Index table, shared-commands callout
    (probe / read / write / subscribe), Driver.Cli.Common infrastructure
    note (what's shared, marginal cost of adding a sixth CLI), typical
    cross-CLI workflows (commissioning, bug reproduction, recipe-write
    validation, byte-order debugging), known gaps that cross-ref the
    per-CLI docs (AB Legacy ab_server upstream gap, S7 PUT/GET enable,
    TwinCAT AMS router, UDT-write refusal), tracking pointer to #249-251.
  - docs/README.md — Client tooling table grows 6 rows (DriverClis parent
    + 5 per-CLI). Also corrects the Client.CLI.md row: it's otopcua-cli,
    not lmxopcua-cli (renamed in #208).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

docs/DriverClis.md

@@ -0,0 +1,90 @@
+# Driver test-client CLIs
+
+Five shell-level ad-hoc validation tools, one per native-protocol driver family.
+Each mirrors the v1 `otopcua-cli` shape (probe / read / write / subscribe) against
+the **same driver** the OtOpcUa server uses — so "does the CLI see it?" and
+"does the server see it?" are the same question.
+
+| CLI | Protocol | Docs |
+|---|---|---|
+| `otopcua-modbus-cli` | Modbus-TCP | [Driver.Modbus.Cli.md](Driver.Modbus.Cli.md) |
+| `otopcua-abcip-cli` | CIP / EtherNet-IP (Logix symbolic) | [Driver.AbCip.Cli.md](Driver.AbCip.Cli.md) |
+| `otopcua-ablegacy-cli` | PCCC (SLC / MicroLogix / PLC-5) | [Driver.AbLegacy.Cli.md](Driver.AbLegacy.Cli.md) |
+| `otopcua-s7-cli` | S7comm / ISO-on-TCP | [Driver.S7.Cli.md](Driver.S7.Cli.md) |
+| `otopcua-twincat-cli` | Beckhoff ADS | [Driver.TwinCAT.Cli.md](Driver.TwinCAT.Cli.md) |
+
+The OPC UA client CLI lives separately and predates this suite —
+see [Client.CLI.md](Client.CLI.md) for `otopcua-cli`.
+
+## Shared commands
+
+Every driver CLI exposes the same four verbs:
+
+- **`probe`** — open a session, read one sentinel tag, print driver health.
+  Fastest "is the device talking?" check.
+- **`read`** — synthesise a one-tag driver config from `--type` / `--address`
+  (or `--tag` / `--symbol`) flags, read once, print the snapshot. No extra
+  config file needed.
+- **`write`** — same shape plus `--value`. Values parse per `--type` using
+  invariant culture. Booleans accept `true` / `false` / `1` / `0` / `yes` /
+  `no` / `on` / `off`. Writes are **non-idempotent by default** — a timeout
+  after the device already applied the write will not auto-retry (plan
+  decisions #44, #45).
+- **`subscribe`** — long-running data-change stream until Ctrl+C. Uses native
+  push where available (TwinCAT ADS notifications) and falls back to polling
+  (`PollGroupEngine`) where the protocol has no push (Modbus, AB, S7).
+
+## Shared infrastructure
+
+All five CLIs depend on `src/ZB.MOM.WW.OtOpcUa.Driver.Cli.Common/`:
+
+- `DriverCommandBase` — `--verbose` + Serilog configuration + the abstract
+  `Timeout` surface every protocol-specific base overrides with its own
+  default.
+- `SnapshotFormatter` — consistent output across every CLI: tag / value /
+  status / source-time / server-time for single reads, a 4-column table for
+  batches, `Write <tag>: 0x... (Name)` for writes, and one line per change
+  event for subscriptions. OPC UA status codes render as `0xXXXXXXXX (Name)`
+  with a shortlist for `Good` / `Bad*` / `Uncertain`; unknown codes fall
+  back to hex.
+
+Writing a sixth CLI (hypothetical Galaxy / FOCAS) costs roughly 150 lines:
+a `{Family}CommandBase` + four thin command classes that hand their flag
+values to the already-shipped driver.
+
+## Typical cross-CLI workflows
+
+- **Commissioning a new device** — `probe` first, then `read` a known-good
+  tag. If the device is up + talking the protocol, both pass; if the tag is
+  wrong you'll see the read fail with a protocol-specific error.
+- **Reproducing a production bug** — `subscribe` to the tag the bug report
+  names, then have the operator run the scenario. You get an HH:mm:ss.fff
+  timeline of exactly when each value changed.
+- **Validating a recipe write** — `write` + `read` back. If the server's
+  write path would have done anything different, the CLI would have too.
+- **Byte-order / word-swap debugging** — `read` with one `--byte-order`,
+  then the other. The plausible result identifies the correct setting
+  for that device family. (Modbus, S7.)
+
+## Known gaps
+
+- **AB Legacy PCCC wire-level** against the ab_server Docker simulator is
+  upstream-broken — see the "Known limitations" section in
+  [Driver.AbLegacy.Cli.md](Driver.AbLegacy.Cli.md). Pointing the CLI at
+  real SLC / MicroLogix / PLC-5 hardware or a RSEmulate 500 golden-box
+  works as expected.
+- **S7 PUT/GET communication** must be enabled in TIA Portal for any
+  S7-1200/1500. See [Driver.S7.Cli.md](Driver.S7.Cli.md).
+- **TwinCAT AMS router** must be reachable (local XAR, standalone Router
+  NuGet, or authorised remote route). See
+  [Driver.TwinCAT.Cli.md](Driver.TwinCAT.Cli.md).
+- **Structure / UDT writes** are refused by the AB CIP + TwinCAT CLIs —
+  whole-UDT writes need a declared member layout that belongs in a real
+  driver config, not a one-shot flag.
+
+## Tracking
+
+Tasks #249 / #250 / #251 shipped the suite. 122 unit tests cumulative
+(16 shared-lib + 106 across the five CLIs) — run
+`dotnet test tests/ZB.MOM.WW.OtOpcUa.Driver.Cli.Common.Tests` +
+`tests/ZB.MOM.WW.OtOpcUa.Driver.*.Cli.Tests` to re-verify.