Allow materializer targets to specify inputs

[elijahbenizzy]

Is your feature request related to a problem? Please describe.
I want to be able to use materializers to create simple ETLs of just moving data from one place to another.

Describe the solution you'd like

A little odd but a good use-case. Specifically, I want something like this:

dr = driver.Builder().build() # empty driver
dr.materialize(
    to.csv(
        id="save_some_input_to_csv",
        dependencies=["some_input"]),
    inputs={
        "some_input" : SOME_DATAFRAME
    }
),

This should form a two-node DAG.

Describe alternatives you've considered
Just writing a function to do this, or doing it outside of Hamilton.

[skrawcz]

Yeah it's because it's empty it doesn't work. It needs a node with that input defined right now to work.

[elijahbenizzy]

Yeah it's because it's empty it doesn't work. It needs a node with that input defined right now to work.

Yep that's why I opened up this issue, cause I think it should work! It's a nice way to use materializers and track them. To avoid pains/encourage early failures we can add a validate_inputs param and default it to true.

[skrawcz]

closing this since it's partially addressed with #911 .