Between redis bloom and cuckoo filters, which is better in terms of performance?

I guess cuckoo filters must have matured quite a bit over the years in terms of adoption.

Cuckoo filters are relatively simple, so no 'maturity process' was required.

That being said, since cuckoo filters introduction in 2014 many improvements have been suggested (and continuously being suggested) including:

• Configurable bucket
• Additive and subtractive cuckoo filter (ASCF)
• Reducing Relocations in Cuckoo Filter
• Tagged Cuckoo Filters
• Optimized Cuckoo Filter (OCF)
• Index-Independent Cuckoo filter (I2CF)
• Leveraging the power of two choices to select the better candidate bucket during insertion

and even

• CFBF: Reducing the Insertion Time of Cuckoo Filters With an Integrated Bloom Filter

Whether each of these methods guarantees better results (insert performance, query performance, memory consumption, etc.) for each use case requires a comparative analysis (I'm not aware of such unbiased research).

As for adoption:

• There are many GitHub repositories implementing cuckoo filter in various languages
• There is a strong academic interest in both theoretical improvements (see above) and applications of cuckoo filters.

So keeping that in mind, which is a better choice among the two in terms of performance as far as Redis implementations are concerned? Is cuckoo filter an obvious choice over bloom given the extra features (like deletion and insertion count)? Are there any trade-offs?

The question you referred to already has excellent answers concerning the performance and the tradeoffs between these two algorithms. It also discusses why performance is not just a single metric (insert performance vs. query performance; average time vs. worst time, etc.). Since Redis implements the data structure described in the original cuckoo filter paper (albeit in a highly optimized way), all issues discussed apply to the Redis implementation as well.

Note that in addition to Bloom and cuckoo filters several additional approximate membership query data structures were suggested, including XOR filters, Ribbon filters, and Binary fuse filters.

Which one is most suitable for each use case requires a non-trivial analysis.