In the sanctions trials, the court focused on sanctionable conduct.

In the trials themselves, the courts decided on the complete lack of evidence of fraud.

The Michigan court rejected the Costantino case in just four days, so I doubt that anyone looked deeply into the evidence.

I don't. The lack of evidence was really obvious.

To know that the affidavits supplied by plaintiffs, purporting fraud, were "rife" with generalization, speculation, hearsay, and a lack of evidentiary basis, all you have to do is read them. Noting that there's a lack of evidentiary basis is as deep as you can look into the evidence in cases like this were none is supplied.

And then ten days later the court rejected an appeal by Costantino because the question of fraud was now moot as the electors had certified the results.

Noting that the lower court had already ruled that there was no evidence of fraud.

Even if it is correct that the Kraken lawyers used fake "evidence" (in which case I agree they should be sanctioned!) that tells us nothing about the hundreds of other claims from all over the country.

It's telling that the ones that were brought before courts all failed.

Hundreds of witnesses are prepared to swear in a court of law, under penalty of perjury, that they witnessed fraud. Not one single one of them has actually been asked to do so.

Who are these hundreds, and why didn't the Kraken lawyers get them instead of the blatant bullshit that they submitted as affidavits?

E.g. videos of election workers unloading ballot boxes at 3am at polling centers when they're supposed to be closed for the night get dismissed as "conjecture"

Can you be more specific with your example? Which polling center are you talking about?

Statistical anomalies like thousands of ballots in a row going to one candidate

That would be a statistical anomaly. What's your example of that?

or results flipping from one candidate to the other overnight when there shouldn't be any vote counting going on

There was a case where results were incorrectly entered into the press system, by typo. There was no case where results flipped overnight, just the reporting of the results were corrected.

In the incredibly rare case that a voting machine is audited

In 2020 44 states required a post election audit. That's better described by "pretty common" than "incredibly rare", isn't it?

and the auditors find that it had been improperly manipulated and data deleted,

Nope, that report was bollocks.

The people of the USA expect #1 holds but the law operates on #2.

The law operates nearer to 1. Transparently false claims of fraud don't mean that the law doesn't investigate election fraud where it might have happened, especially where it might have made a difference.