usb_cdc.serial.read() doc says it reads one byte, but it will read as much as possible

[dhalbert]

Right now the documentation for usb_cdc.serial.read() gives the signature as usb_cdc.serial.read(size = 1). That is, it reads a at most a single byte. This does not match the implementation: it actually tries to read as much as possible, because it uses the standard stream operation read(), which has those semantics.

Pyserial's read() has the same signature and does read only one byte if no argument is given. I think our original intention was to match pyserial.

So rather than change the documentation, we should probably fix the implementation.

[dhalbert]

There are complaints about pyserial's read() behavior:

• Documentation: Serial class's inheritance from RawIOBase/IOBase & inherited methods not documented pyserial/pyserial#285
• Serial.read() does not conform to superclass's io.RawIOBase.read() interface pyserial/pyserial#585
• expected number of bytes on read, waiting for timeout. pyserial/pyserial#425
• read() semantics are broken pyserial/pyserial#318
• serial.read(1) with timeout set returns immediately, when receiving nothing on windows pyserial/pyserial#267

I'm now thinking we may not necessarily want to mimic pyserial so closely. pyserial also oddly released v3.5 to pypi but did not tag the release in github, which shows the last release as v3.4. There have been no commits to master since August 2023.