Add namespace validation for rename operation

[ajantha-bhat]

Fixes #7645

[snazy]

This test is way "too far away" from the changed code. A test needs to be in :nessie-versioned-storage-common-tests - such a test must also verify that the identified-key for CommitValidator works in that case.

[snazy]

Test missing / wrong place.

[ajantha-bhat]

Test missing / wrong place.

I tried adding a test in AbstractNamespaceValidation

@Test
  void RenameWithNonExistingNamespace() throws Exception {
    BranchName branch = BranchName.of("renameWithNonExistingNamespace");
    store().create(branch, Optional.empty());

    Namespace ns1 = Namespace.of("ns");
    Namespace ns2 = Namespace.of("non_existing");
    ContentKey key1 = ContentKey.of(ns1, "table");
    ContentKey key2 = ContentKey.of(ns2, "tbl");

    store()
      .commit(
        branch,
        Optional.empty(),
        fromMessage("create namespace ns"),
        singletonList(
          Put.of(ns1.toContentKey(), ns1)));

    OnRefOnly content = newOnRef("value");
    store()
      .commit(
        branch,
        Optional.empty(),
        fromMessage("create a table"),
        singletonList(Put.of(key1, content)));

    store()
      .commit(
        branch,
        Optional.empty(),
        fromMessage("rename table"),
        asList(Delete.of(key1), Put.of(key2, content)));
  }

But it seems it is not going to the place where I fixed.

I am not very familiar with that part of the code now. It would take too much time for me to understand it now.
Hence, I went with black box level testing. Agree that we should have unit tests for this. But I am not sure whether I
can work on it (or fix it soon).

[ajantha-bhat]

But it seems it is not going to the place where I fixed.

I should fetch the content from the backend to use it for rename. Without that reusing the content object in the test class will be considered as new content as it is not having content id. Fixed now.

[adutra]

Please also add an entry to CHANGELOG.md.

[adutra]

such a test must also verify that the identified-key for CommitValidator works in that case.

Namespace validation happens before the CommitValidator is invoked. The identified key is correctly created, but discarded when the validation error is thrown. I think that's fine?