Fix parquet serialization

[Maxxen]

This PR adds the missing fields in ParquetCopyBind to de/serialization.

Additionally I had to implement a new case in the Deserializer to handle deserialization of unique_ptr's whose element types do not implement ::Deserialize themselves as a static member method, e.g. unique_ptr<unordered_set<idx_t>>.

Also had to format some java class or coequality would fail(?).

Closes https://github.com/duckdblabs/duckdb-internal/issues/572

[Maxxen]

I cannot reproduce this test succeeding locally, will try with an AWS box tomorrow.

[Maxxen]

Introduced a std::sort in the glob branch of the MultiFileReader and it seemed to fix the CSV reader problem on my fork, lets see if it works here too.

[carlopi]

For consistency, should the sort be added also here:

		for (auto &val : ListValue::GetChildren(input)) {
			if (val.IsNull()) {
				throw ParserException("%s reader cannot take NULL input as parameter", name);
			}
			if (val.type().id() != LogicalTypeId::VARCHAR) {
				throw ParserException("%s reader can only take a list of strings as a parameter", name);
			}
			auto glob_files = fs.GlobFiles(StringValue::Get(val), context, options);
+++++++                 std::sort(glob_files.begin(), glob_files.end());
			files.insert(files.end(), glob_files.begin(), glob_files.end());
		}

so that read_csv('some/glob/**') behave the same as read_csv(['some/glob/**']) ?

[Maxxen]

@carlopi Good catch, fixed now!

[carlopi]

I was curious and have given a proper look, LGTM!

[Mytherin]

Thanks! LGTM