Reduce size of Expr

[xitep]

• This whole change is reducing the size of Expr to 112 bytes.
• This appears to be improving the performance of the parser; compared to the main branch #b3e176d, i get this from the benchmarks on my machine (roughly a 7% improvement):

benchmark	main	this-branch
benchmark/sqlparser::select	3.4347 µs	2.8873 µs
benchmark/sqlparser::with_select	17.490 µs	16.946 µs
benchmark/parse_large_statement	5.5956 ms	5.1574 ms
benchmark/format_large_statement	293.73 µs	282.47 µs
word_to_ident/clone_into_ident_100x	2.0907 µs	1.9933 µs
word_to_ident/to_ident_100x	1.6960 µs	1.6310 µs
parse_identifiers/select_100_columns	85.290 µs	78.486 µs
parse_identifiers/select_100_qualified_columns	181.46 µs	165.29 µs

Details

> cargo bench -- --baseline main
...
sqlparser-rs parsing benchmark/sqlparser::select
                        time:   [2.8171 µs 2.8873 µs 2.9611 µs]
                        change: [−20.224% −17.456% −14.693%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high mild
Benchmarking sqlparser-rs parsing benchmark/sqlparser::with_select: Collecting 100 samples in estimated 5.0368 s (308k
sqlparser-rs parsing benchmark/sqlparser::with_select
                        time:   [16.627 µs 16.946 µs 17.299 µs]
                        change: [−8.0130% −5.0889% −2.1257%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high mild
Benchmarking sqlparser-rs parsing benchmark/parse_large_statement: Collecting 100 samples in estimated 5.3783 s (1000 i
sqlparser-rs parsing benchmark/parse_large_statement
                        time:   [5.0584 ms 5.1574 ms 5.2594 ms]
                        change: [−10.235% −7.8314% −5.3302%] (p = 0.00 < 0.05)
                        Performance has improved.
Benchmarking sqlparser-rs parsing benchmark/format_large_statement: Collecting 100 samples in estimated 6.0742 s (20k i
sqlparser-rs parsing benchmark/format_large_statement
                        time:   [276.05 µs 282.47 µs 289.60 µs]
                        change: [−10.490% −8.1724% −5.7397%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 15 outliers among 100 measurements (15.00%)
  14 (14.00%) high mild
  1 (1.00%) high severe

word_to_ident/clone_into_ident_100x
                        time:   [1.9460 µs 1.9933 µs 2.0395 µs]
                        change: [−9.7186% −7.1667% −4.4646%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  3 (3.00%) high mild
  1 (1.00%) high severe
word_to_ident/to_ident_100x
                        time:   [1.5993 µs 1.6310 µs 1.6673 µs]
                        change: [−9.8227% −7.0780% −4.4473%] (p = 0.00 < 0.05)
                        Performance has improved.

parse_identifiers/select_100_columns
                        time:   [76.620 µs 78.486 µs 80.527 µs]
                        change: [−16.581% −13.951% −11.342%] (p = 0.00 < 0.05)
                        Performance has improved.
Benchmarking parse_identifiers/select_100_qualified_columns: Collecting 100 samples in estimated 5.1332 s (30k iteratio
parse_identifiers/select_100_qualified_columns
                        time:   [161.17 µs 165.29 µs 169.87 µs]
                        change: [−14.949% −12.703% −10.289%] (p = 0.00 < 0.05)
                        Performance has improved.

• I can see a similar improve in a client program

[LucaCappelletti94]

If there isn't rush, I can come back with benchmarks in parsing with and without this change against my fuzzing corpus. I am a bit overworked right now, so I don't think I can deliver it earlier than the end of next week.

[xitep]

absolutely no rush, @LucaCappelletti94. (this was just an experiment which surprised me and i thought it's worth sharing.) if you'd manage to run it on some workload of yours it would be a great way to expose or exclude some error on my part.