Mounted filesystems USB write protection

[dhalbert]

• Fixes write-locking problem with USB automount #10510

---

• Ensure that a filesystem visible via USB is read-only to the host if it is read-write to CircuitPython and vice-versa.
• Type-check storage.mount(filesystem=, ...) argument to ensure it is a VfsFat object.
  • Currently it only checks that it implements mount. But elsewhere in the code there are assumptions it is an fs_user_mount_t, which is a `VfsFat.
  • If/when we provide other filesystem types (as MicroPython does), we will relax this. I left the old validation code in as #if 0 for that reason, instead of just deleting this.
• stop_mp() unmounts user-mounted Vfs filesystems on VM shutdown. This used to assume that the user-mounted filesystems all came last. I think that is currently the case, but made this code more robust in case they are intermixed with non-user-mounted ones. Tested on a Fruit Jam with automounted filesystems and on an nRF52840 with a user-mounted SD card.
• Add some clarifying parentheses on a bool expression.
• Add #define PLAT_PRINTF(...) mp_printf(&mp_plat_print, __VA_ARGS__) in circuitpy_mpconfig.h. There are similar macros elsewhere, but only in specific files. I got tired of typing &mp_plat_print. I chose the name for minimal typing for auto-completion.

@tannewt we discussed that open() was not calling filesystem_lock(), etc. I looked at adding these but it's not clear to me that the file objects are properly finalized on VM shutdown, etc., so the locks may not go back to their initial state. Given that this does not actually seem to be a problem right now, I am leaving it alone, to be fixed later when we make more objects have finalizers. This addition is the critical one for fixing the issue

 filesystem_set_writable_by_usb(vfs_fat, readonly);
 filesystem_set_concurrent_write_protection(vfs_fat, true);

[tannewt]

One minor thing.

re: PLAT_PRINT, it's fine to add but note that LLM power complete is much better at completing it and ESP_LOG.

[tannewt]

Thank you!