Remove unused tokens in compilation test

[mooori]

The #[near_bindgen] macro currently ignores tokens after the attribute name, for example #[near_bindgen(this_is_ignored)]. This is due to _attr (which contains these tokens) being ignored:

near-sdk-rs/near-sdk-macros/src/lib.rs

Lines 46 to 47 in 039cc11

	#[proc_macro_attribute]
	pub fn near_bindgen(_attr: TokenStream, item: TokenStream) -> TokenStream {

There's a contract used in compilation test which has tokens after the attribute name: #[near_bindgen(init => new)]. They are removed in this PR.

Context

For near-plugins we're experimenting with passing extra tokens like #[near_bindgen(some_config(value_a, value_b))]. When adding logic to parse these tokens via darling the compilation of the contract with #[near_bindgen(init => new)] fails.

[austinabell]

Sorry if I'm missing the motivation here, but what is the purpose of changing this within our compilation tests? This shouldn't affect what you're building?

Are you suggesting that something within our macros could be changed to allow it to be extended in external libs?

[mooori]

The feature proposed in #952 uses tokens passed after the name of #[near_bindgen]. This comment tries to extend on the motivation and highlights the code that breaks the compilation of cond_compilation.rs.

[mooori]

Assuming init => new is ignored, it might make sense to remove it even if #952 will be rejected. To avoid that someone stumbles upon this in the future.

[austinabell]

Assuming init => new is ignored, it might make sense to remove it even if #952 will be rejected. To avoid that someone stumbles upon this in the future.

Yeah, that makes sense, I'm just making it clear we may use the attributes in the future. #934 is one that proposes this kind of change