[Bug]: No apparent way to build release without debug info

### Which component impacted?

Build

### Is it regression? Good in old configuration?

None

### What happened?

When configuring and building the 22.5.4 release with any of the following CMake invocations, debug info is always included in the final build:

```bash
cmake .. -DBUILD_TYPE=Release
cmake .. -DCMAKE_BUILD_TYPE=Release
cmake .. -DBUILD_TYPE=Release -DCMAKE_BUILD_TYPE=Release
```
Example of the final build flags for a particular source file:
```
-Wreorder -Wsign-promo -Wnon-virtual-dtor -Wno-invalid-offsetof -fvisibility-inlines-hidden -fno-use-cxa-atexit -frtti
-fexceptions -fpermissive -fcheck-new -std=c++1y -std=c++11 -O3 -DNDEBUG -fPIC -Werror -Wall -Winit-self
-Wpointer-arith -Wno-unused -Wno-unknown-pragmas -Wno-comments -Wno-sign-compare -Wno-attributes
-Wno-narrowing -Wno-overflow -Wno-parentheses -Wno-delete-incomplete -Werror=address -Werror=format-security
-Werror=non-virtual-dtor -Werror=return-type -finline-functions -funswitch-loops -fno-short-enums -Wa,--noexecstack
-fno-strict-aliasing -fmessage-length=0 -fvisibility=hidden -fdata-sections -ffunction-sections -Wl,--gc-sections -m64
-DLINUX=1 -DLINUX -DNO_RTTI -DNO_EXCEPTION_HANDLING -DINTEL_NOT_PUBLIC -g -fstack-protector -D_AMD64_
-D__CT__ -D__linux__ -fno-tree-pre -fPIC -Wl,--no-as-needed -O2 -fno-omit-frame-pointer -D_FORTIFY_SOURCE=2
-finline-limit=100 -MD -MT media_driver/CMakeFiles/iHD_drv_video_CODEC.dir/agnostic/gen11/codec/hal/codechal_kernel_hme_g11.cpp.o
-MF CMakeFiles/iHD_drv_video_CODEC.dir/agnostic/gen11/codec/hal/codechal_kernel_hme_g11.cpp.o.d
-o CMakeFiles/iHD_drv_video_CODEC.dir/agnostic/gen11/codec/hal/codechal_kernel_hme_g11.cpp.o
-c /opt/intel/workspace/media-driver/media_driver/agnostic/gen11/codec/hal/codechal_kernel_hme_g11.cpp
```
As can be seen above, the -DCMAKE_BUILD_TYPE=Release does initially set the `O3 -DNDEBUG` flags, but then later on, the `media_compile_flags_linux.cmake` variable `MEDIA_COMPILER_FLAGS_COMMON` globally adds the `-g` flag regardless of build type.

Ultimately this results in a very large shared library binary (~350Mib):
```bash
bloaty /opt/mr/vaapi/lib/dri/iHD_drv_video.so
FILE SIZE VM SIZE
-------------- --------------
67.4% 236Mi 0.0% 0 .debug_info
9.8% 34.2Mi 0.0% 0 .debug_loc
8.0% 28.1Mi 72.3% 28.1Mi .rodata
4.0% 14.1Mi 0.0% 0 .debug_line
3.2% 11.3Mi 0.0% 0 .debug_str
2.1% 7.40Mi 0.0% 0 .debug_ranges
1.9% 6.83Mi 17.6% 6.83Mi .text
1.6% 5.46Mi 0.0% 0 .debug_abbrev
0.5% 1.88Mi 4.8% 1.88Mi .data
0.5% 1.61Mi 0.0% 0 .strtab
0.2% 858Ki 2.2% 858Ki .rela.dyn
0.2% 705Ki 0.0% 0 .symtab
0.2% 682Ki 1.7% 682Ki .eh_frame
0.2% 542Ki 0.0% 0 .debug_aranges
0.1% 298Ki 0.7% 298Ki .data.rel.ro
0.0% 156Ki 0.4% 156Ki .eh_frame_hdr
0.0% 48.1Ki 0.1% 48.1Ki .gcc_except_table
0.0% 40.0Ki 0.1% 31.1Ki [21 Others]
0.0% 18.9Ki 0.0% 18.9Ki .dynstr
0.0% 0 0.0% 9.88Ki .bss
0.0% 8.88Ki 0.0% 8.88Ki .dynsym
100.0% 350Mi 100.0% 38.9Mi TOTAL

```
whereas the version in the Ubuntu 20.04 distribution is only ~7MiB:
```bash
bloaty /usr/lib/x86_64-linux-gnu/dri/iHD_drv_video.so
FILE SIZE VM SIZE
-------------- --------------
48.5% 3.41Mi 48.5% 3.41Mi .text
37.9% 2.67Mi 37.9% 2.67Mi .rodata
4.9% 349Ki 4.9% 349Ki .eh_frame
4.5% 327Ki 4.5% 327Ki .rela.dyn
1.3% 92.7Ki 1.3% 92.7Ki .data.rel.ro
1.1% 76.9Ki 1.1% 76.9Ki .eh_frame_hdr
0.9% 66.9Ki 0.9% 66.9Ki .data
0.3% 18.1Ki 0.3% 18.1Ki .gcc_except_table
0.2% 11.2Ki 0.2% 11.2Ki .dynstr
0.0% 0 0.1% 9.74Ki .bss
0.1% 6.47Ki 0.1% 6.47Ki .dynsym
0.1% 6.01Ki 0.0% 0 [Unmapped]
0.1% 3.98Ki 0.1% 3.98Ki .rela.plt
0.0% 2.68Ki 0.0% 2.35Ki [12 Others]
0.0% 2.67Ki 0.0% 2.67Ki .plt
0.0% 2.66Ki 0.0% 2.66Ki .plt.sec
0.0% 2.19Ki 0.0% 2.19Ki .init_array
0.0% 2.11Ki 0.0% 2.11Ki .fini_array
0.0% 1.81Ki 0.0% 0 [ELF Section Headers]
0.0% 1.52Ki 0.0% 1.52Ki .got
0.0% 688 0.0% 688 .gnu.hash
100.0% 7.03Mi 100.0% 7.03Mi TOTAL
```
After manually deleting the `-g` flag, I was able to get the size down to ~41MiB
```bash
bloaty /opt/mr/vaapi/lib/dri/iHD_drv_video.so
FILE SIZE VM SIZE
-------------- --------------
68.5% 28.1Mi 72.5% 28.1Mi .rodata
16.3% 6.69Mi 17.2% 6.69Mi .text
4.6% 1.88Mi 4.8% 1.88Mi .data
3.9% 1.61Mi 0.0% 0 .strtab
2.0% 858Ki 2.2% 858Ki .rela.dyn
1.7% 710Ki 1.8% 710Ki .eh_frame
1.7% 704Ki 0.0% 0 .symtab
0.7% 298Ki 0.8% 298Ki .data.rel.ro
0.4% 156Ki 0.4% 156Ki .eh_frame_hdr
0.1% 48.0Ki 0.1% 48.0Ki .gcc_except_table
0.0% 18.9Ki 0.0% 18.9Ki .dynstr
0.0% 0 0.0% 9.88Ki .bss
0.0% 8.88Ki 0.0% 8.88Ki .dynsym
0.0% 7.41Ki 0.0% 7.41Ki .rela.plt
0.0% 6.09Ki 0.0% 3.82Ki [14 Others]
0.0% 4.95Ki 0.0% 4.95Ki .plt
0.0% 4.94Ki 0.0% 4.94Ki .plt.sec
0.0% 4.51Ki 0.0% 0 [Unmapped]
0.0% 3.66Ki 0.0% 3.66Ki .init_array
0.0% 3.56Ki 0.0% 3.56Ki .fini_array
0.0% 2.73Ki 0.0% 2.73Ki .got
100.0% 41.1Mi 100.0% 38.8Mi TOTAL
```
I do expect that this latest version will be larger than what is in the distribution, so the above numbers might be what one would expect.

Could someone comment on whether or not I'm missing some aspect around building a Release build without debug info, such that I don't need to manually modify the flags to get the expected binary? Otherwise, might it make sense to move the `-g` flag to the `MEDIA_COMPILER_FLAGS_RELEASEINTERNAL` or `MEDIA_COMPILER_FLAGS_DEBUG` variables?

Edit: Worth noting that gmmlib seems to have the same issue (albeit with a much smaller size) in that `GMMLIB_COMPILER_FLAGS_COMMON` sets `-g` globally as well.

Thanks

### What's the usage scenario when you are seeing the problem?

Others

### What impacted?

_No response_

### Debug Information

Git tags for dependencies:
libva = 2.16.0
gmmlib = intel-gmmlib-22.2.1

### Do you want to contribute a patch to fix the issue?

None