cpu/stm32l5: add initial support + add support for nucleo-l552ze-q board

[aabadie]

Contribution description

This PR adds support to the stm32l5 family and in the meantime to the nucleo-l552ze-q. It's very similar to L4 (regarding clock configuration) but not all bifield defines are available. That's why there's a specific stmclk_l5.c file but the board uses the common l4 clock configuration header.
Only the unsecure mode is available, adding support for TrustedZone and privileged peripheral access is more challenging but could be done in follow-up works.

I was not able to make it work very reliably unless idle thread support is added to the build. My intuition is that there's some sort of race condition at startup but I couldn't find the problem.

• Features tested with success: I2C, UART, GPIO, GPIO irqs, Timer, flashpage, hwrng
• Features missing: RTT, RTC

important: support in OpenOCD is not yet available upstream, you have to build OpenOCD with this patch to program the board.

Testing procedure

• Green Murdock
• Run the compile_and_test_for_board.py script and check the results:

$ BUILD_IN_DOCKER=1 ./dist/tools/compile_and_test_for_board/compile_and_test_for_board.py . nucleo-l552ze-q --jobs=8 --with-test-only
[...]
ERROR:nucleo-l552ze-q:Tests failed: 14
Failures during test:
- [examples/suit_update](examples/suit_update/test.failed)
- [tests/driver_at86rf2xx_aes](tests/driver_at86rf2xx_aes/test.failed)
- [tests/emcute](tests/emcute/test.failed)
- [tests/gnrc_dhcpv6_client](tests/gnrc_dhcpv6_client/test.failed)
- [tests/gnrc_dhcpv6_client_6lbr](tests/gnrc_dhcpv6_client_6lbr/test.failed)
- [tests/gnrc_ipv6_ext](tests/gnrc_ipv6_ext/test.failed)
- [tests/gnrc_ipv6_ext_frag](tests/gnrc_ipv6_ext_frag/test.failed)
- [tests/gnrc_ipv6_ext_opt](tests/gnrc_ipv6_ext_opt/test.failed)
- [tests/gnrc_ipv6_nib_dns](tests/gnrc_ipv6_nib_dns/test.failed)
- [tests/gnrc_rpl_srh](tests/gnrc_rpl_srh/test.failed)
- [tests/gnrc_sock_dns](tests/gnrc_sock_dns/test.failed)
- [tests/gnrc_tcp](tests/gnrc_tcp/test.failed)
- [tests/periph_timer_short_relative_set](tests/periph_timer_short_relative_set/test.failed)
- [tests/pkg_libfixmath_unittests](tests/pkg_libfixmath_unittests/test.failed)

=> No big surprise.

Issues/PRs references

• Made possible by cpu/cortex: add initial support for Cortex-M33 #14367
• Based on cpu/stm32l4: improve clock initialization #15036

[leandrolanzieri]

compile/tests/ztimer_periodic/native:llvm seems to be failing sporadically, it also failed once in #15191.

[benpicco]

Needs a rebase 😉

[aabadie]

rebased

[benpicco]

looks good, just some minor comments

[benpicco]

should those be the same?

[aabadie]

This is not a copy paste issue but indeed both cases could be put in common.

[benpicco]

Looks good. If that's not a copy & paste error in stm32_mem_lengths.mk this should be good to go.
Please squash.

[aabadie]

Squashed and adapted the stm32_mem_lengths.mk.