Fix issues related to introduction of new values in GHEvent

[gsmet]

@bitwiseman I decided to apply the same strategy we used for other things: use Strings for serialization/deserialization and only switch to enums for APIs. It's a bit less optimal but less risky. Another better option would be to have custom Jackson deserializer but that's probably something we need to keep for 2.0.

I also fixed an inconsistency in EnumUtils: one method was user uppercase values and not the other. This is a recipe for disaster.

Better reviewed commit per commit as I kept them semantic.

I'm not exactly sure if it's worth adding tests as we would have to add tests with unknown values for each enum all over the place.

Fixes #1156

[codecov]

Codecov Report

Merging #1159 (54d8fe9) into main (4abf33a) will increase coverage by 0.02%.
The diff coverage is 91.17%.

@@             Coverage Diff              @@
##               main    #1159      +/-   ##
============================================
+ Coverage     73.44%   73.47%   +0.02%     
- Complexity     1834     1839       +5     
============================================
  Files           185      185              
  Lines          6146     6148       +2     
  Branches        367      367              
============================================
+ Hits           4514     4517       +3     
+ Misses         1411     1410       -1     
  Partials        221      221              

Impacted Files	Coverage Δ
src/main/java/org/kohsuke/github/GHHook.java	85.71% <0.00%> (+4.46%)	⬆️
src/main/java/org/kohsuke/github/GHApp.java	72.54% <75.00%> (+1.12%)	⬆️
...ain/java/org/kohsuke/github/GHAppInstallation.java	51.78% <75.00%> (+1.78%)	⬆️
src/main/java/org/kohsuke/github/GHEvent.java	100.00% <100.00%> (ø)
src/main/java/org/kohsuke/github/GHEventInfo.java	93.93% <100.00%> (+8.22%)	⬆️
...in/java/org/kohsuke/github/internal/EnumUtils.java	100.00% <100.00%> (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 4abf33a...54d8fe9. Read the comment docs.

[bitwiseman]

The case sensitivity is intentional. The values coming from GitHub are Pascal-cased and I want to match them specifically or not at all.

Please just add the new events in this PR.

[gsmet]

Mkay. My big concern is that we have two nearly identical methods and the nullable behavior should be the only difference between the two.

I don't mind how we fix it but I think we need to make the methods consistent in one way or another.

I will remove this commit from the PR.

[bitwiseman]

@gsmet
You are completely correct in principle, and I'm almost sure you're right in this case.

I think the reason I chose to do it that way was to match the exiting method's behavior (which was case-sensitive). The events area is not well tested, so I would like to err on the side of caution. Perhaps not it as an issue and link to 38460dc and we can come back to it?

[gsmet]

OK, I removed this commit and did the case change at the call sites.

[hcoles]

Is there anything I can do to help get this merged?

[sathishavm]

I like this change, can this merge?