DC-Sync related changes

[v-morlock]

Hey,
as discussed, this PR contains some dc sync related changes which brought a huge improvement in stability to our system.

The changes are:

• Add methods to be able to send the PDI during transition from PREOP to SAFEOP
• Change how the "first sync0" time is computed, the former approach computed a wrong time for 32bit slaves
• Add a subdevice_mut method (tbh i don't recall why I added it but i use it in my code, I can keep it in my fork if you prefer)
• Add a register_read_ignore_wkc method, this enabled me to check the "next sync0" register, which helped me find the above mentioned issue and to confirm the new approach is indeed working

Please let me know if you want to do some changes, split it up in multiple PR, etc

[v-morlock]

I've also updated the example code to loosely match our production code, I don't have an ethercat slave with me unfortunately, so I'm not able to test it...

[jamwaffles]

Seeing as the DC fix is in #322, could we reduce this PR to just the added methods all_safe_op and request_into_safe_op, and the changes to examples/dc.rs (might want to rebase to get #322 first). I'm not too keen on the subdevice_mut and register_read_ignore_wkc methods and I think we could do the DC example changes in a better way at a later date.

Can you remember why you needed those methods in the first place? Maybe there's a different/better way to get what you need?

[jamwaffles]

P.s. I believe a rebase from main now that #322 is in should fix the tests.

[v-morlock]

Hey there! Sorry for the very late feedback. I wouldn't need the register_read_ignore_wkc method, I think i temporarily used it to debug some dc sync related behavior I cannot recall. However, we're currently using the subdevice_mut method to enable dc sync for individual slaves like that:

        let mut slave = group.subdevice_mut(client, 1)?;
        slave.set_dc_sync(ethercrab::DcSync::Sync0);

which doesn't work without that method - maybe there's a better way?