Update validate_data_hash to return modified hash

[mhashizume]

Puppet::Pops::Lookup::ModuleDataProvider#validate_data_hash is a method that is supposed to prune a data hash of any keys that are not prefixed with a given module's name. However prior to this commit it incorrectly took the data hash, cloned it, operated on the clone, then returned the unchanged original hash.

This commit updates validate_data_hash to instead return the modified hash, and updates the warning message generated when a key is pruned to include the key.

[joshcooper]

Did you want to add a test that this data gets filtered out?

[mhashizume]

Yes, although I've been digging through existing tests and I'm unsure about how to approach this. It would be easy to look at what validate_data_hash returns and make sure that it does not include a given key, but I assume we want to search for and validate that the data has been removed within Puppet. Any pointers?

[joshcooper]

Maybe add a key to a module's hiera.yaml, where the key doesn't match the module's name. Looking at

puppet/spec/unit/pops/lookup/lookup_spec.rb

Lines 57 to 58 in fa213f5

	'common.yaml' => <<-YAML.unindent
	mod::c: mod::c (from module)

You could add an entry like the following:

        'common.yaml' => <<-YAML.unindent
          ....
          other::c: other::c (from module)
        YAML

And then add tests around here

puppet/spec/unit/pops/lookup/lookup_spec.rb

Lines 192 to 194 in fa213f5

	it 'finds data in module layer' do
	expect(Lookup.lookup('mod::c', nil, 'not found', true, nil, invocation)).to eql('mod::c (from module)')
	end

to test that if you do Lookup.lookup('other::c', ...) then it should return nil. And if you lookup an unqualified Lookup.lookup('c', ...) then it should return the mod::c (from module), since the value from other::c should be ignored.

Prior to your changes, I would expect other::c (from) module to be returned in both cases, since it wasn't filtered out.

[mhashizume]

It doesn't seem like unqualified lookups work in tests, both in my PR or in main. I'll dig into it a bit more and see if that's expected, it's a bug, or if there's something wrong with the tests.