Remove use of Object in headers and store header value as rpm tag v…

[dwalluck]

…alue

An attempt at removing some duplicate code between RpmTagValue and RpmInformations.

Improve type safety by always storing a HeaderEntry in the header map instead of raw Object. This required a small API change to call getValue() now to get the actual value, and the stored value was changed to RpmTagValue instead of Object inside the HeaderEntry itself.

I think another possible improvement that is not in this pull request as it requires more code changes is to avoid returning Object from the get() API, and instead return an RpmTagValue and have the caller have to provide asLong() or asString() instead of casting, which may provide better type safety.

[ctron]

Only a small thing.

[ctron]

I am not sure about this change. First of all, it's not aligned with the other (int for example). But also because it won't work with the array variant.

One way to deal with this could be to add a asNumber variant. Which would still need some strategy for the array case I guess.

[dwalluck]

@ctron This code actually came from preexisting code in RpmInformations::asLong that I merged into RpmTagValue::asLong. I don't think I actually changed anything. However, I reverted all changes to this file now except for the absolute minimum, so please look again.

In general, I do not like automatically converting between Integer and Long and I'd like to get rid of it in a different pull request. Java 8 has Integer::toUnsignedLong which we can use for uint32 where needed, but this doesn't actually require changing the data type. We should just change it at the RpmInformations level, but I think we should make that an even higher level, i.e., store an Instant instead of a Long there.