Only define rules for SpecialFunctions if needed

[devmotion]

This PR addresses #319 by only defining rules for SpecialFunctions if SpecialFunctions.ChainRulesCore does not exist, as suggested by @oxinabox in JuliaMath/SpecialFunctions.jl#238 (comment).

Should the same condition be added to the tests? And do you want me to bump the version?

[codecov-io]

Codecov Report

Merging #320 (889ed20) into master (eb10848) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #320   +/-   ##
=======================================
  Coverage   97.46%   97.46%           
=======================================
  Files          18       18           
  Lines         988      988           
=======================================
  Hits          963      963           
  Misses         25       25           

Impacted Files	Coverage Δ
src/ChainRules.jl	100.00% <ø> (ø)

---

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 eb10848...889ed20. Read the comment docs.

[oxinabox]

Yes let's add this to the tests and bump the version also

It will be nice for the tests and we won't see the deprecation warnings.

Can you also open an issue about dropping the SpecialFunctions dependency and cross link that in the code?

[stevengj]

Great! Ideally, I'd like to delay tagging a new SpecialFunctions release until this is merged and tagged in ChainRules, so it would be good if you could do a 0.7.35 release with this in the near future.

[oxinabox]

Will do tomorrow