Mark unstable modules, make macros private

[playfulFence]

Description

closes #2499
potentially closes #2921

Marked the rest of unstable modules.
Makes macros module private (otherwise, it'll cause a huge and noisy diff like crate::before_snippet! -> crate::macros::before_snippet and so on) as it was public only to reexport a couple of things from procmacros.

Adding small "default" branch in esp_hal::init function (caused by instability of watchdog and psram items of Config)

Since rom is also unstable, we need that bad-looking pub use construction for regi2c_write and regi2c_write_mask, which will be removed when rom is stabilized

[MabezDev]

I'll review the code separately soon, but for now I just generated the docs to take a look around.

I believe the following modules still need to be unstable:

• efuse
• asynch
• lp_core

The clock changes are handled in so we don't need to worry about that one: #2899

[playfulFence]

@MabezDev

I believe the following modules still need to be unstable:

• efuse
• asynch
• lp_core

Oops, forgot about them. Then psram too probably 🤔

[playfulFence]

skip-changelog due to changes in esp-hal-embassy which don't require a changelog entry

[bugadani]

What's the reason to make all macros private? Part of that module is public reexports, and we're now reexporting reexported macros?

[playfulFence]

What's the reason to make all macros private? Part of that module is public reexports, and we're now reexporting reexported macros?

After a long fight me and @bugadani decided to leave things as they are because another options (pub unstable macros module) don't seem to be working (also, macros are cursed)

[playfulFence]

UPD: marked entry, handler and ram as unstable only for docs since "truly" marking them as "unstable" leads to a huge amount of errors, not sure what to do with it. It must be used, but esp_riscv_rt nor xtensa_lx_rt, nor procmacros not gonna be 1.0.

Also, it doesn't make sense to stabilize procmacros without stable interrupts.

UPD:
Just

#[instability::unstable]
#[cfg_attr(not(feature = "unstable"), allow(unused))]
pub use procmacros::{handler, ram};

seems to be resolving the issue with procmacros

[bugadani]

entry quite obviously needs to be stable - if you can't write main, how would you use the crate?

[playfulFence]

Exactly, that's what I'm talking about 😅

Which means, we need to either

1. stabilize esp_riscv_rt and xtensa_lx_rt too
2. make users export it separately

[MabezDev]

esp-hal-procmacros is basically tied to esp-hal, the only reason they're in separate a crate is due to compiler limitations, so we'll actually need to make esp-hal-procmacros 1.0 too. I hadn't considered this initially, but this should be fine though, because everything that's in there can also be marked as unstable, and we can do the following:

• doc(hide) the rt crates
  • Optional: we may want to create a doc hidden __macro_impl_details module to consolidate the hidden things that macros might need to touch.
• A new main macro that uses the hidden rt crates, basically #[main] will simply expand to #[arch_rt::entry]
  • because users don't interact with code generated by these macros, we can use them as long as we don't change the behavior of them.
• Add some diagnostics to tell the user to use #[esp_hal_embassy::main] if they try and write an async main.

I suggest we do this in a follow up PR though, as this will involve a bunch more changes. I'll open an issue for this.

[MabezDev]

LGTM, thanks! Some of the smaller things can be tackled outside this PR, but let's get this merged before we hit conflict hell.

[JurajSadel]

Thanks for working on this!