fix(historian): #675 review follow-ups — SupportsReverseEvents capability, no catch-as-unsupported

- IHistorianDataSource.SupportsReverseEvents (default false; SqlServer/QuestDB true; Gateway/Null declare false).
  The node manager checks it BEFORE calling the source: an end-only capped events read becomes reverse only on
  a supporting backend (Gateway, the default, keeps the pre-#675 forward [MinValue, EndTime) paging), and an
  explicit reverse read on a non-supporting backend is BadHistoryOperationUnsupported without a call.
- Removed the broad catch (NotSupportedException) in the events path: a NotSupportedException from inside a
  backend is now logged as a backend fault (Error), not an 'unsupported' Warning.
- Null declares the reverse overload itself; guard test: every production IHistorianDataSource declares both
  the reverse overload and the capability (positive-control scan of all four backends).
- Conformance: capability reported + oversized reverse tie cluster (forward over-fetch reversed == reverse
  read order; capped reverse read stopping inside the cluster flags truncation) on SqlServer + QuestDB.
- docs/Historian.md: end-only capped events read is backward on SqlServer/QuestDB, forward on Gateway.

src/Drivers/ZB.MOM.WW.OtOpcUa.Driver.Historian.SqlServer/SqlServerHistorianDataSource.cs

@@ -311,6 +311,11 @@ public sealed class SqlServerHistorianDataSource : IHistorianDataSource
 
     // ------------------------------------------------------------------------------------------ Events
 
+    /// <inheritdoc />
+    /// <remarks>Reverse events reads are served natively (<c>ORDER BY … DESC</c>, Gitea #675).</remarks>
+    public bool SupportsReverseEvents => true;
+
+
     /// <inheritdoc />
     /// <remarks>
     ///     <c>[start, end)</c> by <c>EventTimeUtc</c> (a reversed window is swapped), ordered by