host-sp-comms: take a second review look at Host Panic/Host Bootfail ereport contents

When merging #2518, there was a lot of commentary on the contents of the MGS messages, but not too much on the contents of the relevant ereports:

https://github.com/oxidecomputer/hubris/blob/d101c3caa7bcbbad8df742ca472771a605e8dbaf/task/host-sp-comms/src/main.rs#L2042-L2082

Looking at this after the fact, @rmustacc pointed a couple of things out, and @jamesmunns noticed:

1. We probably don't need the total_len, an event is probably more uniquely identified by the tuple of (restart_id, seqno), and restart_id did not make it into the ereport is implicitly included in all ereports.
2. HostBootFail renamed seqno to seq, but HostPanic didn't. We should probably make this consistent.
3. We might want to spill a bit more ink on the use of wrapping sequence numbers, and why that is preferable to saturating sequence numbers
4. On Host Bootfail, we may want to consider including some amount of POST code history, as we made available in hubris, from the FPGA

We can either bump the version, or potentially just keep version 0 if we make these changes soon, as I don't believe we've cut a release with these ereports yet. It probably doesn't hurt to rev the version either way though.