stm32/spi: new_blocking_txonly_nosck hangs on v2+

[Dirbaio]

All SPI operations on an SPI created with new_blocking_txonly_nosck hang on chips with SPIv3+ SPIv2+.

It hangs in spin_until_rx_ready.

The only difference vs new_blocking_txonly is we don't configure any pin as SCK. Seems strange the SPI would care, intuitively I'd say SCK should be only an output so the SPI shouldn't care if it goes "nowhere". Perhaps it needs to be "looped back" somehow. Hopefully it's only for the RX part, so we can fix it by actually not doing any RX at all.

[Dirbaio]

fixed in #3028

[Dirbaio]

Reopening since maybe we should try to get it working instead of just disabling it. There's two people that have confirmed it's working for them (in G0 the author of #1345, then another in G4). So i'm very confused why it's always hanging in tests for me...