Implement Yul to LLVM IR compilation benchmarks (#407)

# Description

Closes [#404](https://github.com/paritytech/revive/issues/404)

Adds compilation time benchmarks for:

* Parsing of Yul source code -> AST Object
* Lowering of AST Object -> LLVM IR (unoptimized)

The benchmarks can be run from the root via:

```sh
# Run all benchmarks in the revive-yul crate (parsing + lowering)
make bench-yul
```

HTML reports will be generated under `target/criterion`, and a summary
of the results at
[crates/yul/BENCHMARKS_PARSE_M4PRO.md](https://github.com/paritytech/revive/blob/lj/compilation-benchmarks-yul/crates/yul/BENCHMARKS_PARSE_M4PRO.md)
and
[crates/yul/BENCHMARKS_LOWER_M4PRO.md](https://github.com/paritytech/revive/blob/lj/compilation-benchmarks-yul/crates/yul/BENCHMARKS_LOWER_M4PRO.md)
(currently from running on a Mac M4 Pro).

---------

Co-authored-by: xermicus <cyrill@parity.io>

crates/llvm-context/src/polkavm/context/mod.rs

@@ -256,6 +256,22 @@ impl<'ctx> Context<'ctx> {
         }
     }
 
+    /// Initializes a new dummy LLVM context.
+    ///
+    /// Omits the LLVM module initialization; use this only in tests and benchmarks.
+    pub fn new_dummy(
+        llvm: &'ctx inkwell::context::Context,
+        optimizer_settings: OptimizerSettings,
+    ) -> Self {
+        Self::new(
+            llvm,
+            llvm.create_module("dummy"),
+            Optimizer::new(optimizer_settings),
+            Default::default(),
+            Default::default(),
+        )
+    }
+
     /// Builds the LLVM IR module, returning the build artifacts.
     pub fn build(
         self,