Add ProblemReport section to metadata schema

[ddanielr]

Adds ProblemReport section to the metadata schema to consolidate metadata table prefixes.
This removes hardcoded prefix strings located in various classes.

[dlmarion]

LGTM

[ctubbsii]

I wonder if some of these APIs have a String equivalent, so we don't have to wrap them all with new Text. That could be another slight improvement that would go nicely along with this change (if they are equivalent).

[ddanielr]

I wonder if some of these APIs have a String equivalent, so we don't have to wrap them all with new Text. That could be another slight improvement that would go nicely along with this change (if they are equivalent).

That's a great idea! I just added the changes to remove the Text wrappers.
Mutation already had a CharSequence method signature to use. I didn't think we needed to add a method for Strings, but happy to do so if that's desired.

I also had to modify the Encoding util to support a byte array for full removal of the Hadoop Text wrappers.

This change seemed minimal as Encoding was only being used by ProblemReport and its test class and immediately generated a byte array from the Text object.

[ctubbsii]

One more minor change to eliminate a new Text, but after that, this change looks ready to merge.

[ctubbsii]

This section definition includes the underscore as a mandatory part of the section. That's the most narrowly scoped definition, and completely fine. However, I'm a bit surprised that the section wasn't defined more widely as everything starting with err, as in something like:

      new Section(RESERVED_PREFIX + "err", true, RESERVED_PREFIX + "ers", false);

I am not suggesting changing it... just noting the distinction between the narrow section definition and the wider definition.

[ctubbsii]

One difference is that the new version has the start key inclusive, and the old range has it exclusive. I don't think this matters, since we don't expect any rows to just include the prefix without a tableId after the underscore, so this isn't an important change. I'm just noting it here as a very slight difference in behavior, in case there's a surprise later, and this comment serves to help debug something.

[ctubbsii]

Thanks, @ddanielr . This can be merged once CI checks are finished.