Fixes issue #3880.

And fix timeout value so that it does actually finish with reset_mode=1.

The DMA peripheral is limited to transferring 65535 elements at a time so
in order to send more than that the SPI driver must split the transfers up.

The user must be aware of this limit if they are relying on precise timing
of the entire SPI transfer, because there might be a small delay between
the split transfers.

Fixes issue #3851, and thanks to @kwagyeman for the original fix.

To match pin labels on other NUCLEO 64 boards.

This behaviour of a NULL write C method on a stream that uses the write
adaptor objects is no longer supported.  It was only ever used by the
coverage build for testing the fail path of mp_get_stream_raise().

With this patch objects are only checked that they have the stream protocol
at the start of their use as a stream, and afterwards the efficient
mp_get_stream() helper is used to extract the stream protocol C methods.

The existing mp_get_stream_raise() helper does explicit checks that the
input object is a real pointer object, has a non-NULL stream protocol, and
has the desired stream C method (read/write/ioctl).  In most cases it is
not necessary to do these checks because it is guaranteed that the input
object has the stream protocol and desired C methods.  For example, native
objects that use the stream wrappers (eg mp_stream_readinto_obj) in their
locals dict always have the stream protocol (or else they shouldn't have
these wrappers in their locals dict).

This patch introduces an efficient mp_get_stream() which doesn't do any
checks and just extracts the stream protocol struct.  This should be used
in all cases where the argument object is known to be a stream.  The
existing mp_get_stream_raise() should be used primarily to verify that an
object does have the correct stream protocol methods.

All uses of mp_get_stream_raise() in py/stream.c have been converted to use
mp_get_stream() because the argument is guaranteed to be a proper stream
object.

This patch improves efficiency of stream operations and reduces code size.

If the user button is held down indefinitely (eg unintenionally, or because
the GPIO signal of the user button is connected to some external device)
then it makes sense to end the reset mode cycle with the default mode of
1, which executes code as normal.

Functionality is unchanged.

This was a typo from the very first commit of this file.

It's possible (at least on F4 MCU's) to have RXNE and STOPF set at the same
time during a call to the slave IRQ handler.  In such cases RXNE should be
handled before STOPF so that all bytes are processed before
i2c_slave_process_rx_end() is called.

Due to buffering of outgoing bytes on the I2C bus, detection of a NACK
using the ISR_NACKF flag needs to account for the case where ISR_NACKF
corresponds to the previous-to-previous byte.

These new API functions do not use the cache.

This patch renames the existing SPI flash API functions to reflect the fact
that the go through the cache:

    mp_spiflash_flush -> mp_spiflash_cache_flush
    mp_spiflash_read  -> mp_spiflash_cached_read
    mp_spiflash_write -> mp_spiflash_cached_write

This patch removes the global cache variables from the SPI flash driver and
now requires the user to provide the cache memory themselves, via the SPI
flash configuration struct.  This allows to either have a shared cache for
multiple SPI flash devices (by sharing a mp_spiflash_cache_t struct), or
have a single cache per device (or a mix of these options).

To configure the cache use:

    mp_spiflash_cache_t spi_bdev_cache;

    const mp_spiflash_config_t spiflash_config =
        // any bus options
        .cache = &spi_bdev_cache,
    };

Fixes issue #3859.

This patch changes dupterm to call the native C stream methods on the
connected stream objects, instead of calling the Python readinto/write
methods.  This is much more efficient for native stream objects like UART
and webrepl and doesn't require allocating a special dupterm array.

This change is a minor breaking change from the user's perspective because
dupterm no longer accepts pure user stream objects to duplicate on.  But
with the recent addition of uio.IOBase it is possible to still create such
classes just by inheriting from uio.IOBase, for example:

    import uio, uos

    class MyStream(uio.IOBase):
        def write(self, buf):
            # existing write implementation
        def readinto(self, buf):
            # existing readinto implementation

    uos.dupterm(MyStream())

A port can define QSTR_GLOBAL_DEPENDENCIES to add extra files.

Now that the common module has mbedtls support for both SHA1 and SHA256 it
can now be used on this port.

These are not exported to anyone anyway.

Via the config value MICROPY_PY_UHASHLIB_SHA256.  Default to enabled to
keep backwards compatibility.

Also add default value for the sha1 class, to at least document its
existence.

Makes the code much more readable by reducing the number of #ifdefs and
keeping related functions close.

For consistency with other modules, and to help avoid clashes with the
actual underlying functions that do the hashing (eg
crypto-algorithms/sha256.c:sha256_update).