Don't panic when MultiPublisher is given zero publishers

[imjasonh]

Fixes #215

cc @ian-mi

[codecov-commenter]

Codecov Report

Merging #216 into master will decrease coverage by 0.17%.
The diff coverage is 16.66%.

@@            Coverage Diff             @@
##           master     #216      +/-   ##
==========================================
- Coverage   37.23%   37.05%   -0.18%     
==========================================
  Files          33       33              
  Lines        1402     1414      +12     
==========================================
+ Hits          522      524       +2     
- Misses        790      800      +10     
  Partials       90       90              

Impacted Files	Coverage Δ
pkg/commands/resolver.go	12.57% <0.00%> (-0.81%)	⬇️
pkg/publish/multi.go	60.00% <100.00%> (+10.00%)	⬆️

---

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 01820ce...b890235. Read the comment docs.

[jonjohnsonjr]

This ignores -P or -B flags, and always uses -P behavior.

Hm... what if we always append the default publisher and just added push as a default publisher Option and guarded all this with a boolean:

https://github.com/google/ko/blob/dc30efcfd8b6f6a8aa381c715390013f9c1fa23b/pkg/publish/default.go#L128-L151

Then that keeps everything in sync and we don't have to worry about naming logic drift, plus we can just return an error from multipublisher if len(publishers) == 0 and it's pretty easy to follow the logic.

[jonjohnsonjr]

~/go/src/github.com/google/ko ((b890235...)) $ ko publish -B --push=false ./cmd/ko/
2020/10/06 09:44:44 Using base gcr.io/distroless/static:nonroot for github.com/google/ko/cmd/ko
2020/10/06 09:44:46 Building github.com/google/ko/cmd/ko for linux/amd64
gcr.io/jonjohnson-test/ko/ko@sha256:77582412aa5b7919ff07951ebdac05fd145463c802ddf254bca3065292c3059f

Dope.