What to do about untrustworthy jaccard --> ANI estimates?

[bluegenes]

When estimating ANI from Jaccard (only for scaled jaccard searches via sourmash search), we can run into cases where the error around the Jaccard estimate itself is too high for the Jaccard --> ANI conversion to be trustworthy.

We currently notify with a warning and store ani_untrustworthy in SearchResult.

WARNING: Error on Jaccard distance point estimate is too high (0.0010).

We could:

1. add ani_untrustworthy column to Jaccard search output (though it is irrelevant for any other search/ANI)
2. zero out the ANI estimate (""/None probably better than 0, as 0 is a valid ANI result)

[ctb]

I like zeroing it out. That, combined with the warning, should be enough.