[clang] [HLSL] Collect explicit resource binding information (PR #111203)

[Helena Kotas via cfe-commits]

hekota wrote:

> If there's no test coverage that exercises this case then it might be better to remove the if guard so that it doesn't accidentally mask some other problem we weren't expecting.

The `if (!DBI)` case gets hit when parsing line 112 here where the user defined type `Eg12` has two resources classes of the same type: https://github.com/llvm/llvm-project/blob/5e9166e02ab65d42efba014f2adc59c42b097ddc/clang/test/SemaHLSL/resource_binding_attr_error_udt.hlsl#L112
(even though it is also testing duplicate register bindings).

https://github.com/llvm/llvm-project/pull/111203