Create WelchConfig object

[haberdashPI]

This allows welch_pgram to use a predefined set of configuration parameters in an object named WelchConfig, rather than passing these configuration values by keyword argument. FFT plans and intermediate buffers are preconfigured in WelchConfig making repeated calls that use the same config object faster.

This also sets MTConfig and WelchConfig to be children of the abstract type AbstractPGramConfig

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1204607356620460

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (be1a084) 97.56% compared to head (3f356c0) 97.58%.
Report is 6 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #502      +/-   ##
==========================================
+ Coverage   97.56%   97.58%   +0.01%     
==========================================
  Files          18       18              
  Lines        3124     3147      +23     
==========================================
+ Hits         3048     3071      +23     
  Misses         76       76              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[martinholters]

Should we introduce

periodogram(s::AbstractVector{T}, config::WelchConfig) where T<:Number = welch_pgram(s, config)
periodogram(signal::AbstractVector, config::MTConfig) = mt_pgram(signal, config)

? I'm not sufficiently familiar with these functions to tell whether this makes sense, but if it doesn't, the common supertype AbstractPGramConfig would be questionable as well.

[martinholters]

Testing welch_pgram! would be good to get it covered as well.

[wheeheee]

On the topic of abstract supertypes, Filter doesn't seem to be doing anything

[martinholters]

On the topic of abstract supertypes, Filter doesn't seem to be doing anything

True. I wouldn't mind it being deleted if it bothers you.

[martinholters]

Generally LGTM.
The suggestions concerning the default for buffer are just that - suggestions.

As I don't know this code too well, another approval before merge would be welcome.

[wheeheee]

With so many configs lying around, would it be good to also add configs for Periodogram and Periodogram2? I mean, not in this PR, of course.

[wheeheee]

I'm also not too familiar with this code, but I looked through it and I think it's good for merging.
Also FWIW I ran tests for SignalAnalysis.jl master (which uses welch_pgram) with this PR after fixing up digitalfilter and compat there and they passed.
Merging soon, if there aren't further objections.