ffm PWM channel other than 0 fail to export after reboot.

[taartspi]

There are two statements that use chip rather than channel in PwmFFMHardware

Reboot and repeated attempt to create a PWM on a channel > 0 will then work after two fails. Happens as the first attempts each progress further through the initialization code.

I will continue investigation of Smoke test fail along with these fixes.

[taartspi]

In working on this as linuxfs sees less problems I looked at that code. Well, now i remember when I completed linuxfs PWM for the Pi5 RP1, I added a delay when exporting so the new contexts and files could become persisted in storage prior to accessing them, Something similar likely is needed in ffm

[taartspi]

PR 564 fixes this

[FDelporte]

Looking into the pull request, it's not clear if the delay is needed for the test implementation or required in all PWM use cases? Should the delay be part of the provider?