Should CDriver::surface_movement[iZone] also depend on iInst?

[maxaehle]

The Address Sanitizer (see #2212) reports the following memory leak for the serial testcase hb_rans_preconditioning:

Direct leak of 10992 byte(s) in 6 object(s) allocated from:
    #0 0x13c232e in operator new(unsigned long) (/home/aehle/SU2/install/bin/SU2_CFD+0x13c232e) (BuildId: 8e780270f481da176f9f289b39b954c6eef56689)
    #1 0x14a7288 in CDriver::PreprocessDynamicMesh(CConfig*, CGeometry**, CSolver***, CIteration*, CVolumetricMovement*&, CSurfaceMovement*&) const /home/aehle/SU2/build/../SU2_CFD/src/drivers/CDriver.cpp:2392:24
    #2 0x148c1d0 in CDriver::CDriver(char*, unsigned short, int, bool) /home/aehle/SU2/build/../SU2_CFD/src/drivers/CDriver.cpp:211:7
    #3 0x14c3707 in CFluidDriver::CFluidDriver(char*, unsigned short, int) /home/aehle/SU2/build/../SU2_CFD/src/drivers/CDriver.cpp:2878:98
    #4 0x14c9347 in CHBDriver::CHBDriver(char*, unsigned short, int) /home/aehle/SU2/build/../SU2_CFD/src/drivers/CDriver.cpp:3111:33
    #5 0x13c67fa in main /home/aehle/SU2/build/../SU2_CFD/src/SU2_CFD.cpp:143:18
    #6 0x14f88dabfd84 in __libc_start_main (/lib64/libc.so.6+0x3ad84) (BuildId: c52dbe785ce4223dafe0b6ec0f6e25ea0c8e14f4)

The problem is that CDriver::surface_movement[iZone] is constructed in CDriver::PreprocessDynamicMesh, which is called from CDriver::CDriver in a loop over all iZone and iInst. So if nInst[iZone] > 1, CDriver::surface_movement[iZone] is initialized multiple times. This could be fixed by storing separate CSurfaceMovements for all values of iInst, i.e. CDriver::surface_movement[iZone][iInst], but I wonder if that is correct.

Setup: Ubuntu 22.04 with Clang++ 16.0.6, MPI disabled, SU2 approximately at recent develop branch at f9f7a01.