* Start working on fixing doxygen warnings

* Upgrade doxyfile to Doxygen 1.9.1

* Clean up some unmatched groups and add some exclusions

* Fix docs for rtos::ThisThread not showing up

* Fix up KV store docs

* Fix incorrect @files, some more incorrect groups

* Fix lots of broken links

* Start fixing mbedtls groups

* Fix more mbedtls docs

* Fix remaining mbedtls files with missing or incorrect grouping

* Fix a huge amount of PSA docs errors

* Fix a few more mbedtls docs errors

* Fix __cplusplus not defined for docs, clean up call in callback typedef, fix some docs errors

* Fix some broken doc links in HAL headers

* Fix more PSA doc errors

* Fix some doxygen errors in networking code

* Fix lots of network related doc errors

* Try and fix some CI failures

* Use Doxygen Action instead of doxygen in Mbed container to get the right version

* Try removing docs check

* Add first part of doxygen action

* Add next part of doxygen action

* Fix yaml syntax error

* Add in license check fix

* Try using verbose scancode

* Fix some more docs issues, workaround scancode-toolkit#3179

* Clean up the LoRaRadio get_state() function on each implementation.  Also fix some other compile and doc errors.

* Fix all remaining doxygen errors

* Fix escaping, for realsies this time

* Remove Doxygen unsupported config options, fix some more get_status() declarations

* Fix some UNITTESTS build errors

* Fix one more compile error

* Another unittest error

Currently `TDBStore::offset_in_erase_unit()` and
`TDBStore::check_erase_before_write()` loop through erase units
one-by-one, until the entire range is covered. This is very inefficient
when the erase size is tiny, e.g. one-byte on a non-flash device for
which we use program as erase.

This commit reworks the algorithms, based on the fact that a block
device can erase or program as many units as needed in one go.