Health thresholds are based (delta) by comparing against some previous build.
When, for an existing job, you add a new checker, or make an existing checker more strict, you typically cause all the health thresholds to be violated.
In large, legacy code bases, its never realistic to expect to burn those down by any appreciable amount in any finite time.
But, you can however, expect to attempt to contain growth of the newly identified and tracked technical debt.
Is it possible to have some sort of UI option, command, whatever to say, "accept this build as the baseline to compare against"
The current scenarios to effect this without the option are fairly ugly.
Unless I'm missing something, the options as I see them are
1. delete the job and recreate it whenever you add or make more strict a checker. This will establish the baseline at build 0 but loose all the trending and historical data
2. raise/disable the health threshold gate (e.g from 0 new normal to 10'000 new normal), run a build so you have a new 'stable' base, then drop the limit back down.