FLASHFILE is now a generated file when doing `make all`.
This prepares also for when flashers will use `FLASHFILE` as a file to
be flashed.

It currently still needs the hack below for openocd and edbg.

This also fixes the issue when building 'riotboot' in docker that was
being built with the host toolchain.

Use the new variable instead of the composed name.

fix unexpected behavior:
reference #9949
reference #10294

fix redefinition for AUTO_INIT_LED0

The last line should be a float lower than 1 so 0.XXX.
So expect it to match this.

This should prevent issues when float printing does not work.

Even if using `%f` the `printf_float` module was not used.

When running the test on `samr21-xpro` and `arduino-mega2560` the float
is not printed correctly.

 * samr21-xpro ` false positive rate.`
 * arduino-mega2560 `? false positive rate.`

As the arduino-mega2560 does not handle printf_float use
`fmt/print_float`.

The output should be flushed before using fmt/print functions if
available as they do not use `printf` buffer.

Python strings for regular expression should be declared as `raw` so `\`
is not re-interpreted.

This fixes `flake8` warning.

sizeof(ip6_addr_t) > 16 depending on compilation flags, possibly reading
more data than available.

Fix by using the actual address' sizeof for memcpy.

When running the test on `arduino-mega2560` printing the float failed
and was printed as ` ?`.

    Calculated  ? bits of entropy from 10000 samples.

Replace using `printf` floating point printing by using `fmt/print_float`.
Now the test succeeds on `arduino-mega2560`.

As `print_float` does not buffer and is used with `printf` the output
should be flushed before calling it if `fflush` is available.

cpp check is reporting the following error

    error (shiftTooManyBitsSigned):
    Shifting signed 32-bit value by 31 bits is undefined behaviour

Fix by ensuring the `1` is an `uint32_t` before shifting.

Improved readability of if statements. This issue was pointed out by cppcheck.

This driver currently only supports the SPI connected DS3234, and only
for initializing 1 Hz square wave output on the SQW pin, nothing else.

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

Updated the test to use NETOPT_MAX_DPU_SIZE instead of the deprecated
NETOPT_MAX_PACKET_SIZE

After being cleared, the memory can be `0xFF` for `atmega` platforms for
example.

Fix the test to take EEPROM_CLEAR_BYTE into account.

`strcmp` was used to check that the 4 bytes were empty but was not
testing what it should as an empty string is `0` bytes long.

The whole must be verified.

Use `memcmp` where it is obvious it is not doing a string comparison.

Casting to `const char *` is not necessary as input, the type are
already the right ones.

It could not be before as `result` was with the wrong type.

Results type should only be an array of char, not an array of pointers.

To make the CI happy ;-).

Several parts were inspired by a driver written by:
  George Psimenos (gp7g14@soton.ac.uk)
  Steffen Robertz (steffen.robertz@rwth-aachen.de)

Add a test case to the crypto tests testing the chacha20poly1305
algorithm with the test vectors from rfc 8439.

Add positive and negative tests for data bits, stop bits and parity:

- 7E1, 7O1
- 8E1, 8O1
- 8N2

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>

Add command mode that will be used like this:

mode <dev> <data bits> <parity> <stop bits>

This command must be called after init otherwise
the UART won't be fully initialized.

Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>