Improve multiple updates SQL

[roji]

This PR is mainly about synchronizing our update SQL to be the same as what we use in the TechEmpower implementation, I'll be submitting changes to make them 100% the same in parallel (see TechEmpower/FrameworkBenchmarks#8005).

One run gave me +3.52%, but the variance in updates is huge (another run showed a slight decrease). So this isn't about perf, but rather synchronization and also having simpler SQL.

db	baseline	new_update_sql
CPU Usage (%)	66	66	0.00%
Cores usage (%)	1,850	1,836	-0.76%
Working Set (MB)	57	57	0.00%
Build Time (ms)	1,499	1,450	-3.27%
Start Time (ms)	443	420	-5.19%
Published Size (KB)	523,872	523,872	0.00%

application	baseline	new_update_sql
CPU Usage (%)	63	62	-1.59%
Cores usage (%)	1,757	1,749	-0.46%
Working Set (MB)	570	547	-4.04%
Private Memory (MB)	1,229	1,181	-3.91%
Build Time (ms)	3,044	3,153	+3.58%
Start Time (ms)	1,619	1,676	+3.52%
Published Size (KB)	96,483	96,483	0.00%
Symbols Size (KB)	45	45	0.00%
.NET Core SDK Version	8.0.100-preview.3.23159.6	8.0.100-preview.3.23159.6

load	baseline	new_update_sql
CPU Usage (%)	4	4	0.00%
Cores usage (%)	112	114	+1.79%
Working Set (MB)	48	48	0.00%
Private Memory (MB)	363	363	0.00%
Start Time (ms)	0	0
First Request (ms)	110	112	+1.82%
Requests/sec	25,636	26,537	+3.52%
Requests	387,079	400,695	+3.52%
Mean latency (ms)	23.58	22.69	-3.77%
Max latency (ms)	186.13	168.41	-9.52%
Bad responses	0	0
Socket errors	0	0
Read throughput (MB/s)	18.52	19.17	+3.51%
Latency 50th (ms)	18.17	17.60	-3.14%
Latency 75th (ms)	25.53	24.68	-3.33%
Latency 90th (ms)	37.52	35.67	-4.93%
Latency 99th (ms)	127.31	119.09	-6.46%

[DamianEdwards]

Is this one ready to go?

[roji]

This was already done in 83993ea, closing.