Current in few places we are using hashing algorithms, like for example in Rows::hash() function.

The definition of this method looks like that:

Rows::hash(string $algorithm = 'xxh128', bool $binary = false, array $options = []) : string

Another exampole would be Flow\ETL\Function\Hash or GroupBy (that is using hashing internally)

What we should do instead is replace it with already exising abstraction Flow\ETL\Hash\Algorith which is just a simple interface in all of those places and simply pass NativePHPHash instance as a default.

So for example Rows could look like this:

- Rows::hash(string $algorithm = 'xxh128', bool $binary = false, array $options = []) : string
+ Rows::hash(Algorith $algorithm = new NativePHPHash()) : string

If any of the adapters (like for example elasticsearch) are also using some form o hashing, they should as well be converted to relay on Algorithm interface.