gen_gewisano_local: Generate local GewiSano Weights
In ooelrich/oscbvar: Stuff for Bayesian vector autoregressions

Generates weights according to Geweke & Amisano 2011/2012, also known as linear predicition pools. Takes a data frame of similarities and uses only "similar" data when finding optimal weights.

1	gen_gewisano_local(data, start_t, weight_df, pratig = FALSE)

`data`	Data set of atomic predictions.
`start_t`	Which timepoint to start generating weights for. Obs! This is based on the variable t in the data, not the row number.
`weight_df`	Data fram generated by caliper_relevance_new.
`pratig`	If true, the function prints a bunch of information when run. It also returns a data frame that contains the weights at each time, so pratig has to be FALSE when running from one of the meta-functions that generate predictions. This is because the object returned changes from a single data frame to a list with the original data frame returned as the first object, and a matrix of weights as the second object.