Post init promisc config

[kivkiv12345]

Added csp_zmqhub_<add/remove>_filters() to change promisc settings after init.

Tested with Valgrind without warnings or errors. I didn't add a mutex, as it didn't appear to be necessary.

This is currently used by the following PyCSH commit: spaceinventor/PyCSH@0e398e4#diff-a1457fd5997277b0ab808d1e18fdd8b8d874b88ead8ebd1b33ee4e6e4ff8fdb9
Which is implemented in pycsh_core here: spaceinventor/pycsh_core@c07939a#diff-b8f67f3336f931c16d8f7da067ee582289b8a735040cb5a4fbc2735deaf72e96R228

I got tired of having to close CSH because someone added promisc to my interfaces :)

[jeanbaptistelab]

That netmask assignment is new, isn't it ? Actually, so is the addr assignment.
I don't think it's an issue, I just noticed this wasn't there before. I've looked at the other csp_zmqhub_init_* functions and they do set the addr but not the netmask.
Whether it's something we should streamline or not, I don't know...

[kivkiv12345]

For completeness I'll answer here as well.
Previously we made this assignment in CSH after the csp_zmqhub_init_filter2 call, which seems pretty dumb since every possible caller will have to make the same assignment (from arguments which are given to csp_zmqhub_init_filter2 no less).
For this PR to work, I need the fields to be assigned before calling the promisc functions.
So it was a good excuse to improve some wonky call semantics.

[kivkiv12345]

Actually @jeanbaptistelab, I know this PR has already been merged, but I've been wondering if it would make sense to make a more generic way to call the promisc/unpromisc functions for CAN as well (once they are implemented).
For now it will be up to the user to determine the type of the interface (likely by checking which .nexthop is used), which is not very nice.