and to a mixing studio, you ONLY send WAV (or AIFF) tracks, preferably at 24bit, normally at 44.1 kHz (or 48 kHz); higher frequencies (96 kHz or 88.1) only weigh down the project (but this could be debated here)
heaven forbid we send them the project with MIDI tracks; that would mean they would need to have the same SW and samples that we use, and THEY would have to do the work of converting them into audio files (with the risk that they might mess things up)
this is if you have clear ideas and manage to do the production on your own; if instead you want to rely on a studio also for the choice of samples, recording of other tracks, etc... then perhaps it might also make sense to pass the project, and in that case, the type of sequencer used can be important...
as far as managing orchestral libraries is concerned, as Geppino says, Cubase/Nuendo also handle the various articulations quite easily
It depends on what kind of mix is being done... In a mixing studio oriented towards cinema where they also do dubbing, it is possible that project files are also sent. They even attach the specifications of the plug-ins to be used for a certain voice, including an attachment for the settings of each individual plugin. The owners of the studios were telling me that for the dubbing of some cartoons, they would attach a kilometer of documentation on the voice plugins. If the mix is purely musical, then what you say is true: stems are sent. It becomes untrue again for sound design in commercials, where it is more convenient to send clean projects.
The choice of 44.1 or 48 for a project is not made at random. 44.1 and multiples are the standard for music; 48 and multiples are the standard for cinema. Sampling at multiples of these frequencies is, however, very convenient. Frequencies that are not in the audible band can still produce effects in the audible band, so it is always advisable to sample at very high frequencies, remembering that by sampling at a certain frequency x, we put ourselves in a position to capture frequencies up to (almost) x/2. Obviously, it is always preferable to have 24 bits of dynamic range.
