feat: Introduce msg_limit for application subnets in payload builder

[stiegerc]

After inducting a stream slice, the number of signals in the outgoing stream is equal to number_of_signals_after_gcing + number_of_messages_inducted. So if we want to cap the number of signals in a stream at 50k, we can induct no more than 50k - number_of_signals_after_gcing number of messages.

Gc'ing signals is done using begin in the stream slice, so number_of_signals_after_gcing = outgoing_stream_signals_end - incoming_slice_begin; the signals_end of the outgoing stream is given by message_index of ExpectedIndices.

One way or another this means we now have a cap on how many messages we can extract from the slice pool in any case. Since there already is a mechanism for limiting the number of messages called msg_limit it seems logical to modify this accordingly. Note that this is not handed in from outside but rather initialized by the function get_msg_limit() in get_xnet_payload_impl() and validate_slice(). To include what I laid out out above in get_msg_limit(), I need the signals_end of the outgoing stream and the begin of the stream slice we want to extract.

In validate_slice() both are available, in get_xnet_payload_impl() the begin of the stream slice is not. But msg_limit is only passed down to a place where it is.

What this PR does is move the call to get_msg_limit() further down to where all the variables are available. All I have to do for that is pass a reference to the state rather than the msg_limit.