Remove get_buildtype_args function

[bruchar1]

This is a first step to make buildtype a true alias of debug and optimization options.

See #10808.

Relates to:

• buildtype not derived from debug and optimization options, with surprising results #11645
• buildtype does not get deduced from debug and optimization options #12096
• override_options: ['buildtype=release'] not working, possible regression. #5920
• ‘optimization’ affects buildtype #5814
• Base \MD[d] and \MT[d] off of 'debug' instead of 'buildtype' #8220
• There is no way to access what vscrt is finally used in the build #8493
• Using -Dbuildtype and -Doptimization + -Ddebug is not redundant despite the warning. #9540
• Meson needs -Dbuildtype=custom to change the optimisation & debug values #10487
• buildtype=minsize won't minimize the binary with Visual Studio #12265
• Add b_ndebug=auto because b_ndebug=if-release is misleading #8308
• [Feature Request] Add 'default' to optimization and debug #8214
• [Docs] Difference between buildtype=plain and buildtype=custom #7194
• Setting b_vscrt via meson setup --reconfigure has no effect on the generated build.ninja #11732

[jpakkane]

Getting everything optimization and debug info related away from buildtype is a good thing. In fact I thought that is what we already did but apparently not.

I don't think we can deprecate buildtype for the reasons listed in the referred issue (it carries semantic meaning) so things like get_option('buildtype') have to keep working, possibly forever.

[dcbaker]

I like buildtype in some ways, but in others I don't. Really the only thing that I see being really useful is debug vs non-debug, and we already have a flag.

Could we at least start with deprecating -Dbuildtype and get_option('buildtype')? We'd need some patches from @bonzini's b_pie rework for the former, but the latter would be two lines of code in func_get_option(). That would at least get the ball rolling for the full removal whenever we're ready to have a 2.0 backwards incompatible change.

[jpakkane]

Really the only thing that I see being really useful is debug vs non-debug, and we already have a flag

There needs to be some way of specifying the intent of "this build will be used for X". Trying to guess it from optimization flags or something is not reliable and if we don't provide this, every project will add their own which are different and won't work together.

[jpakkane]

The question what to do with buildtype is unrelated to this MR, which is merely a refactoring. It should not block this.