Fix endpoint checking; Clean up safe mode printing

[dhalbert]

Fixes #4746.
Fixes #4752 (STM32 HID support).

• Endpoint count checking was off by one.
• CIRCUITPY_USB_{HID,MIDI,MSC}_ENABLED_DEFAULT default values were not set conditionally.
• Clean up print_safe_mode_message() to simplify logic, reduce size, and divide up safe-mode reasons more accurately.
• Redo logic of printing startup messages in run_code_py() to not print the safe-mode message twice, clean up newlines, and be more consistent about whitespace.
• Shortened and rephrased some messages slightly.
  EDIT:
• Count in and out endpoints and check against provided limits. ESP32-S2 allows only 5 IN endpoints, so this enforces that limit
• Turn on MIDI and HID on some low-endpoint boards, but have them disabled. The user could disable something else (like the REPL) to add them.
• Add more information to documentation.

It appears to me that a translate() message with a \n in it is actually printed as a \r\n by serial_write_compressed(). I'm not sure if that's done during printing or during compression. But a bare serial_write("\n") does not print a \r. So I added \r as necessary in a few places.

[jepler]

Looks OK, no testing performed.

[Neradoc]

It did not fix #4746 in my tests, both cases with issues now do a reset loop.

[dhalbert]

It did not fix #4746 in my tests, both cases with issues now do a reset loop.

Thanks, because of debugging issues I was testing with an STM board, but will try with an ESP32-S2.

still broken

[dhalbert]

@Neradoc It turns out the ESP32-S2 endpoint situation is idiosyncratic. From the datasheet:

• Endpoint number 0 always present (bi-directional, consisting of EP0 IN and EP0 OUT)
• Six additional endpoints (endpoint numbers 1 to 6), configurable as IN or OUT
• Maximum of five IN endpoints concurrently active at any time (including EP0 IN)

All the other chips have symmetric pairs of IN/OUT endpoints, but the ESP32-S2 does not. I am confused why we can support HID at all, because there don't appear to be enough singular endpoints. It's late and I will look in the morning. In any case, I will need to restrict the choices. It will be possible to add more HID devices, but I think the CDC data channel is not possible. (EDIT: misunderstanding on my part due to the way the datasheet was written. I looked at the code.)

So there are six additional endpoint pairs, but there can be only up to 5 IN endpoints among EP0 and the six. We need EP0, and then we need four more: one IN for MSC, one IN for HID, two IN per CDC (control and data). So there are not enough for the secondary CDC unless MSC or HID is turned off. There could be MIDI. I'll have to decide how dynamic I can make this, There are static allocations in TinyUSB for each device.

[dhalbert]

Note edit in previous comment.

[tannewt]

Looks fine to me. It might be good to try an update the translations so that the ones aren't lost about button presses.

I'll be making more changes; dismissing just so it won't be merged.

[dhalbert]

Added:

• Count in and out endpoints and check against provided limits. ESP32-S2 allows only 5 IN endpoints, so this enforces that limit.
• Turn on MIDI and HID on some low-endpoint boards, but have them disabled. The user could disable something else (like the REPL) to add them.
• Add more information to documentation.

make translate seems to have fixed up some of the messages that would otherwise have been lost due to rewriting. I will check after merging and address missing ones from weblate if I can.

[tannewt]

Looks good to me! Thanks!