In house SEQ explorer application/website

[jonborchardt]

The allennlp team wants to be able to see how alterations to models affect the results of SEQ pipelines. The ask is to have a tool that looks kind of like other demos that they can choose a seq model, change options, (like using random values instead of existing token vectors) and play with them to see results. This could also be a report of x number of known seed phrases to get back new completions for inspection.