CUDA lacks support for external flags and especially for -ccbin (aka --compiler-bindir) flag

[ghost]

Describe the bug
A clear and concise description of what the bug is.

To Reproduce
Please include your meson.build files, preferably as a minimal toy example showing the issue.
You may need to create simple source code files (don't include private/proprietary code).

Expected behavior
A clear and concise description of what you expected to happen.

system parameters

• Is this a cross build or just a plain native build (for the same computer)?
• what operating system (e.g. MacOS Catalina, Windows 10, CentOS 8.0, Ubuntu 18.04, etc.)
• what Python version are you using e.g. 3.8.0
• what meson --version
• what ninja --version if it's a Ninja build

[ghost]

NVCC passes off some of its work to the local c/cc compiler, but it doesn't support all compilers or all versions of the compilers it does support. On some systems, like OpenSuse, its common for GCC to be the latest available (v10.2 as I write this) while the installed CUDA only supports earlier GCC versions. NVCC has a -ccbin flag whereby one can specify which compiler to use, so its possible to install an alternative compiler to use.

However, the existing cuda language support ignores all external flags during sanity checks, so there is no way to set up ccbin in such a way that cuda passes the sanity checks when starting.

How to Reproduce:
Use a version of NVCC that doesn't support the current compiler, and try to get it to work.

Expected Behavior:
I would expect that either having an environment variable of CUFLAGS or, better yet, a cuda_ccbin argument to the language setup would allow someone to use NVCC with an alternative compiler.