feat: v3 support tracking issue

[twuebi]

Feature Request / Improvement

Hi all, let's use this issue to track our progress on adding v3 support to iceberg-go. We can base ourselves on this pyiceberg issue: apache/iceberg-python#1818

• New DataTypes

  • UnknownType feat(types): Support Unknown Type for v3 table spec #605
  • Timestamp(Tz)NanoType feat(types) add support for TimestampNano(Tz) #594
  • VariantType
  • Geometry/Geography

• Default values

• row lineage

• Multi-arg transforms

  • Reading of the metadata has been added in feat(table): Table Metadata V3 support #585
  • Supporting multi-arg transforms still needs to happen

• Row-level lineage

• Deletion vectors

  • Read support
  • Write support

[Shreyas220]

Hey @twuebi @zeroshade I’d like to take up implementing Deletion Vectors.

I’m thinking of breaking this down into these PRs:

1. Puffin file writer/reader
2. DV reader (the manifest file stores the exact byte offset/length of the bitmap so can directly read that )

2.5 Positional Delete Manifest builder
Needed before DV writer to properly create Positional DeleteFile entries as deletion vectors are just content = 1 (POSITION_DELETES) and file format as PUFFIN . (also thinking of slipping the Iceberg v2 positional delete Parquet writer here if possible.)

3. DV writer

Let me know if this order makes sense or if something should be grouped differently.