Fail Maven build on compiler warnings; remove some warning suppressions

[Marcono1234]

Currently this enables all compiler warnings (-Xlint:all); I am not sure if that is a good idea. If you want we can also restrict it to specific warnings instead. For example warnings for serial seem to be redundant here and varargs seems to not be very helpful. I am also a bit afraid that the supported and detected warnings depend on the JDK version being used (instead of the --release version being specified).

It also looks like javac and Eclipse IDE disagree regarding when a warning needs to be suppressed. For some of the required suppressions here Eclipse is telling me "Unnecessary @ SuppressWarnings".

Supersedes #1374

Please let me know what you think.

[Marcono1234]

I am also a bit afraid that the supported and detected warnings depend on the JDK version being used (instead of the --release version being specified).

Yes, that can be seen with the failed CIFuzz run:

Warning:  COMPILATION WARNING : 
[INFO] -------------------------------------------------------------
Warning:  source value 7 is obsolete and will be removed in a future release
Warning:  target value 7 is obsolete and will be removed in a future release
Warning: [WARNING] To suppress warnings about obsolete options, use -Xlint:-options.
[INFO] 3 warnings 
[INFO] -------------------------------------------------------------
[INFO] -------------------------------------------------------------
Error:  COMPILATION ERROR : 
[INFO] -------------------------------------------------------------
Error:  warnings found and -Werror specified

[eamonnmcmanus]

I'm OK with this if you can fix the CIFuzz problem. I suspect we may find that failOnWarning is more trouble than it's worth but it won't be hard to back just that part out if so.

[Marcono1234]

Have removed this comment because it is a bit misleading (at least at the current line where this is placed). It appears this was added in 2011 after experiments with Field.setInt and similar. Though I am not sure if that code is part of the Git history. Additionally since currently Gson only supports Object values, primitives would always be boxed. It would also have to be verified if these 5% difference still applies after now 11 years.

[Marcono1234]

Have excluded the options warnings which were causing the failure for CIFuzz, and have also tried to improve the placement for some @SuppressWarnings annotations. I hope that is alright.

Though I agree, if failOnWarning is causing too much issues we could remove it again.

[eamonnmcmanus]

This is great, thanks!