Don't touch ActiveTime in preSeal/postUnseal

[ncabatoff]

If a consumer of the sys/leader api relies on ActiveTime, they can be misled by our current behaviour, whereby ActiveTime gets set to "now" during postUnseal and to the zero time during preSeal. We sometimes preSeal/postUnseal while remaining the active node. Instead, only set these when we actually change active role.

[github-actions]

CI Results:
All Go tests succeeded! ✅

[github-actions]

Build Results:
All builds succeeded! ✅

[raskchanky]

[banks]

LGTM. I can't think of any case where we'd rely specifically on the old behavior here and I suspect this restores the intended behaviour (certainly the expected behavior) which was presumably broken when we introduced ways to seal and unseal without standing down which I guess only applied to Integrated Storage?

Thanks for noticing the issue!

[ncabatoff]

LGTM. I can't think of any case where we'd rely specifically on the old behavior here and I suspect this restores the intended behaviour (certainly the expected behavior) which was presumably broken when we introduced ways to seal and unseal without standing down which I guess only applied to Integrated Storage?

Thanks for noticing the issue!

No, I think the preSeal/postUnseal-while-remaining-active behaviour goes back at least as far as replication - i.e. not IS-specific. But yeah, I think this is a strict improvement over the current behaviour.