Consider flipping peek_output_buffer() naming convention in next major release

[HadrienG2]

As discussed in #28 , in an ideal world, peek_output_buffer() would be called output_buffer(), and output_buffer() would be called output_buffer_mut(). And the input_buffer() API would work similarly. But that's a breaking API change, so ideally it would be done gradually.

• First, publish a minor release that adds input_buffer_mut() and output_buffer_mut() methods and throws a deprecation warning when calling input_buffer() or output_buffer() (which are now aliases to these methods), with a warning that their semantics are going to change to read-only in the future.
• Next, a few months to a year later, publish a major release which changes input_buffer() and output_buffer() to read-only access, removes the deprecation warning, and deprecates the peek_output_buffer() method.
• Finally, a few months to a year later, publish another major release which removes the peek_output_buffer() method.

[HadrienG2]

The first step (add _mut accessors and deprecate current non-mut accessor names with mutable access) has been taken in v8.1.0 on February 2, 2025.

In the next major release, I'll take the second step of flipping the non-mut accessors to read-only access (which will complete the migration for the producer side) and deprecating peek_output_buffer().

[HadrienG2]

The second step (making non-mut accessors read-only) has been taken in PR #47, which is destined to eventually become v9.0.0.