Is a Differential probe -a la micsig 1300v- going to give the accuracy i need?

[Fungus]

Getting back to the point: Differential is the correct way to go. The bandwidth of those Micsig probes is 100Mhz and the gain accuracy is 2%, ie. better than most 'scopes input gain amplifier accuracy.

On differential probes, common mode rejection is usually more of a limitation than frequency response.  It does not matter when the probe displays high frequency content when it is meaningless.

Depending on the model, the Micsig probes seems to fare quite well.

There's a thread about it: https://www.eevblog.com/forum/testgear/new-eevblog-hv-probe/