feat: Parquet reader builder supports building multiple ranges to read

[evenyag]

I hereby agree to the terms of the GreptimeDB CLA.

Refer to a related PR or issue link (optional)

• Support parallel scan in mito engine #2806

What's changed and what's your intention?

This PR defines the FileRange struct for parquet files and implements a method build_file_ranges() to build ranges from a parquet file. A FileRange contains a range of rows to read from a parquet file. We can read different ranges in parallel later. Now a FileRange is a row group exactly.

pub struct FileRange {}

impl FileRange {
    pub(crate) async fn reader(&self) -> Result<RowGroupReader> {}
}

To reuse code, this PR

• Refactors the ParquetReader. Now it adds a RowGroupReader to read Batches from a row group. The ParquetReader invokes the RowGroupReader to read the parquet file.
• Defines a struct FileRangeContext for all ranges of the same parquet file. This PR also moves the precise_filter() method to the context as the context contains all inputs the method needs.

Now the builder uses a method build_reader_input() to construct the FileRangeContext and row groups to read. Both build() and build_file_ranges can reuse this method.

impl ParquetReaderBuilder {
    pub async fn build(&self) -> Result<ParquetReader> {
        let (context, row_groups) = self.build_reader_input().await?;
        ParquetReader::new(context, row_groups).await
    }

    pub async fn build_file_ranges(&self, file_ranges: &mut Vec<FileRange>) -> Result<()> {}

    async fn build_reader_input(&self) -> Result<(FileRangeContextRef, RowGroupMap)> {}
}

This PR also fixes some remaining issues and improves the ReadFormat helper struct.

• It passes the projection while constructing the struct so it could
  • compute the projection_indices in advance
  • compute the field_id_to_projected_index in advance. Then convert_record_batch() doesn't require &mut self. This is necessary for the FileRangeContext as we have to share the ReadFormat
• It wraps the SimpleFilterEvaluator into a SimpleFilterContext. The context gets the column info in advance, from the expected region metadata. This ensures the context can use the correct column id to find the column in the ReadFormat

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR does not require documentation updates.

[codecov]

Codecov Report

Attention: Patch coverage is 88.25397% with 37 lines in your changes are missing coverage. Please review.

Project coverage is 85.32%. Comparing base (154f561) to head (53396e1).
Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3841      +/-   ##
==========================================
- Coverage   85.60%   85.32%   -0.28%     
==========================================
  Files         952      953       +1     
  Lines      163079   163431     +352     
==========================================
- Hits       139596   139451     -145     
- Misses      23483    23980     +497     

[fengjiachun]

LGTM

[v0y4g3r]

LGTM