Add more fill variants

[dbalagansky]

I'm not an expert in numpy, so this was borrowed from SO: https://stackoverflow.com/a/54508024.

I probably should rename this from "previous" to "similar" (or maybe some other better word?), as this thing covers the case when first point in resulting series could be NaN with backwards fill in addition to the standard forward fill.

Fixes #1266

[tobias-urdin]

Please add testing to show functional behaviour, a release note would be good to, see [1] for using reno to add a release note.

[1] https://docs.openstack.org/reno/latest/user/usage.html#creating-new-release-notes

[tobias-urdin]

@Mergifyio rebase

[mergify]

rebase

❌ Unable to rebase: user tobias-urdin is unknown.

Please make sure tobias-urdin has logged in Mergify dashboard.

[tobias-urdin]

@Mergifyio rebase

[mergify]

rebase

✅ Branch has been successfully rebased

[chungg]

sort of surprised we didn't have this but you should update doc/source/rest.j2 as well.

not blocker but:

• similar is a odd keyword. not sure if there's existing keyword we can copy from other libs? bidirectional seems more accurate but maybe too long?
• would be good to have forward and backward as well. i suspect forward is the most common option.

[dbalagansky]

• would be good to have forward and backward as well. i suspect forward is the most common option.

I've been thinking about these for some time and the problem with plain backward and forward fill variants is that, given this series: [NaN, NaN, 1, 2, NaN, 4, 5, NaN, NaN]:

• doing just forward fill would result in: [NaN, NaN, 1, 2, 2, 4, 5, 5, 5]
• doing just backward fill would result in: [1, 1, 1, 2, 4, 4, 5, NaN, NaN]

So, as you can see, in some cases it wouldn't be enough to do forward or backward fill to produce fully populated series (without NaN's) and, maybe, it would be a good idea to:

• add something like full_ffill for doing forward than backward (for overwriting first two NaN's in the example above after normal forward fill) and same thing with full_bfill: backward than forward fill. This way there would be no need for similar or bidirectional fill variants as the one that is implemented in this PR is just full_ffill one
• add ffill and bfill, which would leave NaN's in the resulting series alone

So maybe implementing one of the above (or both?) would be a better option in the end?

[chungg]

* add something like `full_ffill` for doing `forward` than `backward` (for overwriting first two `NaN`'s in the example above after normal `forward` fill) and same thing with `full_bfill`: `backward` than `forward` fill. This way there would be no need for `similar` or `bidirectional` fill variants as the one that is implemented in this PR is just `full_ffill` one

* add `ffill` and `bfill`, which would leave `NaN`'s in the resulting series alone

yeah, this makes sense to me. i didn't even consider the differences in order of your full fill solution. i'm ok with the naming. don't have anything better to suggest

[dbalagansky]

This should be ready for review, hope I didn't miss any points, but feel free to point them out :)

One more thing that came to my mind during the final tests was that this probably should interact with resampling down (e.g., doing aggregates resample mean 5 on a metric with measurements every minute) by creating additional measurements and filling them, but that sound like a separate PR.

[chungg]

lgtm. minor wording update