PG-286: Fix query buffer overflow management.

If pgsm_overflow_target is ON (default, 1) and the query buffer
overflows, we now dump the buffer and keep track of how many times
pg_stat_monitor changed bucket since that.

If an overflow happen again before pg_stat_monitor cycle through
pgsm_max_buckets buckets (default 10), then we don't dump the buffer
again, but instead report an error, this ensures that only one dump file
of size pgsm_query_shared_buffer will be in disk at any time, avoiding
slowing down queries to the pg_stat_monitor view.

As soon as pg_stat_monitor cycles through all buckets, we remove the
dump file and reset the counter (pgss->n_bucket_cycles).

pg_stat_monitor.h

@@ -306,6 +306,21 @@ typedef struct pgssSharedState
 	pg_atomic_uint64	prev_bucket_usec;
 	uint64				bucket_entry[MAX_BUCKETS];
 	char				bucket_start_time[MAX_BUCKETS][60];   	/* start time of the bucket */
+	LWLock				*errors_lock;		/* protects errors hashtable search/modification */
+	/*
+	 * These variables are used when pgsm_overflow_target is ON.
+	 *
+	 * overflow is set to true when the query buffer overflows.
+	 *
+	 * n_bucket_cycles counts the number of times we changed bucket
+	 * since the query buffer overflowed. When it reaches pgsm_max_buckets
+	 * we remove the dump file, also reset the counter.
+	 *
+	 * This allows us to avoid having a large file on disk that would also
+	 * slowdown queries to the pg_stat_monitor view.
+	 */
+	bool				overflow;
+	size_t				n_bucket_cycles;
 } pgssSharedState;
 
 #define ResetSharedState(x) \