Code duplication in AEGIS implementations

[asterized]

Hi, I've noticed that in the AEGIS-256 and AEGIS-128L implementations there seems to be some code duplication of the functions aegis128l_update and aegis256_update (e.g. in aegis128l_soft.c, aegis128l_aesni.c). Would these be better grouped under the common header or under each implementation?

[jedisct1]

These files are compiled with different CPU features. Implementations of the same primitive can also use completely different representations.

The "soft" implementations are for example likely to be eventually replaced with something derived from https://github.com/aegis-aead/aegis-bitsliced .

And functions are small and simple, so keeping them close to where they are used actually makes things easier to understand and maintain.