(1) As the first option, you should define it as INTERIOR since the beginning when you construct the computational domain. When you import it into Fluent, you don't need to deal with it.
(2) In the second option, you don't need to creat your geometry part by part. If you draw your geometry part by part, sometimes you need to define INTERIOR at unnecessary interfaces. Since your geometry is simple, you can draw as one unit and use submap function in meshing. By doing like this, you can avoid inside interfaces and you will not need to define INTERIOR.
(3) Please also wait for the suggestions from other experienced researchers.
Liquid-vapor interface conditions are coupled heat and mass transfer conditions in general. These are not easy in FLUENT as per my knowledge. However if you have no vaporisation and if you want to calculate back pressure, then I think you need not have a outlet-1 boundary at all. The edge defined for outlet-1, should be a common part of the faces it occupies. Or it can be defined as interior.
I'm not entirely sure how you want to model your case. From the figure you attached it seems that outlet 1 boundary should be an interior zone. As if you generate a single body mesh, pressure inlet condition in the inlet, pressure outlet condition in outlet 2 boundaries and the fluid domain (including the outlet 1 zone) will be calculated by fluent. If that's the case you should use interior type at outlet 1. Since the pressure in the outlet 1 will be calculated by fluent, the pressure may not be equal to 3 bar.
If you want to specify the pressure at outlet 1 boundary then there is no need for you to model all geometry at once. Since you define the pressure at the intersection of inlet channel and the larger chamber, there are no interaction between them. If that's the case you generete two cases. One for inlet channel pressure inlet =4 bar and pressure outlet = 3 bar, one for large chamber with pressure inlet = 3 bar and pressure outlet = 0.