drivers/at24mac: fix dependency

[benpicco]

Contribution description

The at24mac driver does not depend on at24cxxx, EEPROM and ID functionality are separate.

Testing procedure

tests/driver_at24mac still works

2020-11-16 12:15:07,922 # main(): This is RIOT! (Version: 2021.01-devel-903-g96afd-drivers/at24mac-dep)
2020-11-16 12:15:07,926 # EUI-48: fc c2 3d 0d 2d 1f
2020-11-16 12:15:07,932 # ID-128: 0a 70 08 00 64 10 04 81 1c cd a0 00 a0 00 00 00
2020-11-16 12:15:07,933 # [SUCCESS]

Issues/PRs references

[aabadie]

accidentally introduced by #14386

It was already like this when #14386 was opened.

[benpicco]

Huh this is odd.
I now remember that I added this dependency because I thought the linker would just throw out all at24cxxx_ functions if they are not used, so only having to include one module is more convenient.

Yet this saves 988 bytes of ROM on tests/driver_at24mac compared to master.

[benpicco]

ah this is because at24cxxx pulls in xtimer and that will cause changes to core functionality even if xtimer is not called by the application.

So yea, I don't know if this makes a difference in practice.
The dependency is only for convenience, should we still have it?

[benpicco]

at24c% pulls in at24cxxx so it makes sense that the user should never have to manually include at24cxxx.
In that sense having the dependency for at24mac is consistent - and any application that uses MAC addresses will also use xtimer.

I think this can be closed.