bpo-46841: Use *inline* caching for BINARY_OP

[brandtbucher]

Here's a sort-of-rough first pass at implementing this idea. BINARY_OP is easy to start with, since it only uses its cache for the adaptive counter member. Initial perf testing doesn't show any impact, which I think is good (it means that the basic idea works well, even when the cache isn't being used).

Related changes:

• All of the dis utilities grow a show_caches parameter, defaulting to False.
• wordcode_helpers.h has been moved into compile.c.

https://bugs.python.org/issue46841

[markshannon]

Maybe don't document this until we are sure it's what we want?

[brandtbucher]

I'd prefer to keep it, and change it if/when dis changes (which is easy enough). That way we don't forget to document it.

[markshannon]

OK. Once we've started doing this we will need to complete it before the beta release anyway.

[markshannon]

Looks good in general.
I've commented on https://bugs.python.org/issue46841 about what interface we want to add to the dis module.

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @brandtbucher for commit 1aa079c 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

[markshannon]

👍
Let the process of inlining the caches commence