Fixing locking issues for document type saves. (#15854)

* Added  ExecuteNonQuery(DbCommand command) on database to ensure we call OnExecutingCommand and OnExecutedCommand when executing DbCommands

* Added Cache Instructions lock, to avoid deadlocks

* Optimized read locks for nucache when only one content type is rebuilt

* Optimized the SqlServer locks, so only one command is executed (and thereby roundtrip) per lock instead of two

* Avoid breaking changes

* Cosmetic changes

* Take locks if everything is rebuild

* Use same lock in scopes, to avoid potential deadlocks between the two

* Use eager locks in PublishedSnapshotService.cs

* Added timeouts to some of the application locks

* Revert "Use eager locks in PublishedSnapshotService.cs"

This reverts commit 01873aae978ffa6e6686d253e482c493715e3a96.

* Revert "Added Cache Instructions lock, to avoid deadlocks"

This reverts commit e3fca7c12a804bb32ca1156b8abd42a957e9dc21.

* Use single readlock call to lock many

* Use eager locks for reads

* Eager write locks

* Ignore test of lazy locks

* Unique timeout exception messages

---------

Co-authored-by: kjac <kja@umbraco.dk>

src/Umbraco.PublishedCache.NuCache/ContentStore.cs

@@ -27,6 +27,8 @@ namespace Umbraco.Cms.Infrastructure.PublishedCache;
 /// </remarks>
 public class ContentStore
 {
+    private static readonly TimeSpan _monitorTimeout = TimeSpan.FromSeconds(30);
+
     // TODO: collection trigger (ok for now)
     // see SnapDictionary notes
     private const long CollectMinGenDelta = 8;
@@ -330,7 +332,12 @@ public class ContentStore
             throw new InvalidOperationException("Recursive locks not allowed");
         }
 
-        Monitor.Enter(_wlocko, ref lockInfo.Taken);
+        Monitor.TryEnter(_wlocko, _monitorTimeout, ref lockInfo.Taken);
+
+        if (Monitor.IsEntered(_wlocko) is false)
+        {
+            throw new TimeoutException("Could not enter monitor before timeout in content store");
+        }
 
         lock (_rlocko)
         {