Make Signals immutable, restructure annotations, clean up read and write methods

[OTDE]

Okay, this ended up being a bigger PR than I expected -- would love to hear thoughts on any suggested changes or things I'd need to walk back here.

Extreme makeover: EDF Edition

EDF.jl

• Signals are now immutable
• AnnotationsList and RecordAnnotation have been restructured into a Pair{RecordAnnotation, TimestampAnnotation} to better reflect that the first annotation in each data record is both required and structured differently
• RecordAnnotation and TimestampAnnotation both inherit from AbstractAnnotation, allowing for dispatch during EDF.write
• Header renamed to FileHeader as multiple header records exist in an EDF file
• Since multiple signals contain headers, each is given a representation, allowing for clearer write and read functions

I/O

• Functions are more finely divided into discrete sections
• Variables are given more descriptive names
• writing functions now dispatch on new annotation types

Tests

• Pretty much updated to work with the new structs somewhat.

There's a lot here, so I decided this would probably be a good stopping point before I added anything else (and tbh I probably should have requested input a bit sooner), but I'm hoping this PR will make transitioning to an Mmap solution much easier, which, in turn, could make future EDF endeavors a bit more friendly on the memory.

Things that still need to be added, should I get the go-ahead

• Docs
• NEW tests

[codecov-io]

Codecov Report

Merging #23 into master will decrease coverage by 1.96%. The diff coverage is 96.68%.

@@           Coverage Diff           @@
##           master   #23      +/-   ##
=======================================
- Coverage   98.96%   97%   -1.97%     
=======================================
  Files           3     3              
  Lines         194   200       +6     
=======================================
+ Hits          192   194       +2     
- Misses          2     6       +4

Impacted Files	Coverage Δ
src/EDF.jl	100% <100%> (ø)	:arrow_up:
src/read.jl	95.53% <94.59%> (-2.76%)	:arrow_down:
src/write.jl	98.59% <98.52%> (-1.41%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update e3a341b...7770b60. Read the comment docs.

[OTDE]

OOF. Remembering how much the me of two months ago needed to reread YASGuide. These'll be up in a sec.

Do you know whether this has an impact on performance, e.g. @benchmark EDF.read("file.edf")

I don't, unfortunately. I could set something up in the next tagged release so that we have a way to check this, if that works!

[ararslan]

Should be super quick and simple to do. Check out master, run

using BenchmarkTools
using EDF
@benchmark EDF.read("some_edf_file.edf")

Then check out this branch and run the same thing and report both results. You could use one or more of the EDF files checked in as test cases here to benchmark.

[OTDE]

Old:

julia> @benchmark EDF.read("test/data/test.edf")
BenchmarkTools.Trial: 
  memory estimate:  3.33 MiB
  allocs estimate:  8071
  --------------
  minimum time:     937.574 μs (0.00% GC)
  median time:      962.879 μs (0.00% GC)
  mean time:        1.068 ms (7.88% GC)
  maximum time:     5.179 ms (81.16% GC)
  --------------
  samples:          4677
  evals/sample:     1

New:

@benchmark EDF.read("test/data/test.edf")
BenchmarkTools.Trial: 
  memory estimate:  3.60 MiB
  allocs estimate:  15404
  --------------
  minimum time:     1.347 ms (0.00% GC)
  median time:      1.413 ms (0.00% GC)
  mean time:        1.618 ms (9.79% GC)
  maximum time:     5.273 ms (65.65% GC)
  --------------
  samples:          3087
  evals/sample:     1

Looks like performance is being affected.

[OTDE]

After latest test:

julia> @benchmark EDF.read("test/data/test.edf")
BenchmarkTools.Trial: 
  memory estimate:  3.33 MiB
  allocs estimate:  6819
  --------------
  minimum time:     884.232 μs (0.00% GC)
  median time:      913.414 μs (0.00% GC)
  mean time:        1.014 ms (8.20% GC)
  maximum time:     2.654 ms (54.53% GC)
  --------------
  samples:          4924
  evals/sample:     1