The restrained check used a fixture gauge, while the released measurement used the drawing datums; correction followed the first failed released check, but only the final accepted number survives in the sheet.

That is a useful thing to explain to whoever writes the new instructions, Aaron Chan, because otherwise the operator sees weld complete followed by accepted and has to discover the missing judgement themselves

I'd keep the failure note even without its number. It tells you the process included a correction decision. It doesn't tell you the size of the error or prove that every assembly needed the same work.

Bruno, did you recover any before-correction readings on yours, or did you need new samples?

New samples. The old gauge survived, but that did not recreate the missing readings or identify which components had gone into each old bracket.

That is the trap I'd avoid here too. Owning the measuring tool can feel like owning the evidence. It isn't, though it may help define a proper current-part comparison.

Fabrication confirms the current gusset revision differs from the old one, so we are defining a current-assembly study rather than transferring the old correction allowance into the UR5e estimate as a known quantity.

Will that study include the untouched released assembly, Aaron Chan, before anyone applies the familiar correction and leaves you with another very tidy final result?

And price the inspection and possible correction as unresolved work until that scope is agreed. Unknown is inconvenient, but it isn't the same as no work needed.