v2
You should be able to supply the source as a float. That's probably the most common use case, rather than choosing it via a string input. So I've changed the name of the sole exported function in V1 from f_emaFlowBias to f_emaFlowBiasSource()
And updated the f_emaFlowBias() function to accept a float as the source. You can just choose what you pass in, which is much more logical.
With apologies to the very few people who might already be using this new library, I think this is the logical way round.