Support reading V3 tables #1540

Feature Request / Improvement

Currently, we raise an exception when reading V3 tables. As a first step, we should be able to read V3 tables.

An overview can be found here: https://iceberg.apache.org/spec/#version-3

This, for example, entails correctly deserializing the source-ids, raising an exception when there is more than one source-id, and then pushing it into source-id. This way we can break up the V3 support in PyIceberg into smaller issues, and also unblock other things like the support for deletion vectors: #1516 We should also make sure that we are not yet able to produce V3 tables.