feat: collector automatically merge and align multiple collect() called with different schema

[skalwaghe-56]

Implemented automatic schema merging for collectors that:

• Unions fields from multiple collect() calls instead of requiring identical schemas
• Fills missing fields with null values during execution
• Maintains consistent field ordering (alphabetically sorted by field name)
• Preserves auto-generated UUID fields across merged schemas

Closes #428.

[georgeh0]

Hi @skalwaghe-56, sorry for the late reply! I missed it somehow.

[georgeh0]

I don't quite like sorting fields. I think we want to preserve the original order if possible. We only need to add one restriction - if the same field appears on both, they must have consistent in ordering (otherwise raise an error). Then we can merge them without changing the order.

One possible merging approach is (pseudo code, only to show the gist):

let mut output_fields = vec![];
let next_field_id_1 = 0;
let next_field_id_2 = 0;
for (idx, field) in schema2.iter().enumerate() {
  if Some(idx1) = field index in schema1 {
    if (idx1 < next_field_id_1) {
      api_bail!("order mismatch...");
    }
    output_fields.extend(schema1.fields[next_field_id_1..idx1]);
    output_fields.extend(schema2.fields[next_field_id_2..idx]);
    output_fields.push(merged field);
    next_field_id_1 = idx1 + 1;
    next_field_id_2 = idx + 1;
  } else if field is uuid {
    // For UUID, emit it immediately to make sure it still appears first
    ....
  }
}
next_field_id_1.extend(schema1.fields[next_field_id_1..]);
next_field_id_2.extend(schema2.fields[next_field_id_2..]);