This does not change the existing behavior, because it's 0 by
default. This allows us to increase the optimization level when
DEBUG=TRUE.

ERF's radiation test problem does not compile with
`USE_CUDA=TRUE DEBUG=TRUE`. It fails at link time with
```
/usr/lib/gcc/x86_64-linux-gnu/13/../../../x86_64-linux-gnu/Scrt1.o: in function `_start':
(.text+0x21): relocation truncated to fit: R_X86_64_GOTPCRELX against symbol `__libc_start_main@@GLIBC_2.34' defined in .text section in /lib/x86_64-linux-gnu/libc.so.6
/usr/lib/gcc/x86_64-linux-gnu/13/../../../x86_64-linux-gnu/Scrt1.o:(.eh_frame+0x20): relocation truncated to fit: R_X86_64_PC32 against `.text'
/usr/lib/gcc/x86_64-linux-gnu/13/../../../x86_64-linux-gnu/crti.o: in function `_init':
(.init+0xb): relocation truncated to fit: R_X86_64_REX_GOTPCRELX against undefined symbol `__gmon_start__'
/usr/local/cuda/lib64/libcudart_static.a(cudart_static.o): in function `libcudart_static_a3bf49b59a3e5c73e23c76622c3ce0951724bc73':
cuda_runtime_api.cpp:(.text.startup+0x19): relocation truncated to fit: R_X86_64_PC32 against `.data'
cuda_runtime_api.cpp:(.text.startup+0x20): relocation truncated to fit: R_X86_64_PC32 against `.data'
cuda_runtime_api.cpp:(.text.startup+0x27): relocation truncated to fit: R_X86_64_PC32 against `.data'
cuda_runtime_api.cpp:(.text.startup+0x3d): relocation truncated to fit: R_X86_64_PC32 against `.data'
/usr/local/cuda/lib64/libcudart_static.a(cudart_static.o): in function `libcudart_static_7bd86af72cd50dfe63513852abf557d8ffc48b76':
cudart_global.cpp:(.text.startup+0xbb): relocation truncated to fit: R_X86_64_PC32 against `.bss'
cudart_global.cpp:(.text.startup+0xcb): relocation truncated to fit: R_X86_64_PC32 against symbol `__dso_handle' defined in .data.rel.local section in /usr/lib/gcc/x86_64-linux-gnu/13/crtbeginS.o
cudart_global.cpp:(.text.startup+0xd2): relocation truncated to fit: R_X86_64_PC32 against `.bss'
/usr/local/cuda/lib64/libcudart_static.a(cudart_static.o):(.eh_frame+0x4b18): additional relocation overflows omitted from the output
ERF3d.gnu.DEBUG.MPI.CUDA.ex: PC-relative offset overflow in PLT entry for `mlock@@GLIBC_2.2.5'
collect2: error: ld returned 1 exit status
make: *** [../../../Submodules/AMReX/Tools/GNUMake/Make.rules:64: ERF3d.gnu.DEBUG.MPI.CUDA.ex] Error 1
```

With `USE_CUDA=TRUE DEBUG=TRUE DEBUG_OPT_LEVEL=1`, it compiles.