performance of contains is not good

[DavideViolante]

Hello,
I have a huge dataset of data on influx for at least 4 years (data every 2 minutes).
We found out that our queries using contains are way slower than the one using a simple OR operator.

Fast, some ms:

from(bucket: "fybra")
    |> range(start: -7d)
    |> filter(fn: (r) => r["device"] == "123456" or r["device"] == "789012", ...)

Super slow (30+ seconds or timeout):

from(bucket: "fybra")
    |> range(start: -7d)
    |> filter(fn: (r) => contains(value: r.device, set: ["123456", "789012", ...]))

How is this even possible?

Might be related or duplicate of: #3546