feat: Add selectivity metrics to FilterExec

[2010YOUY01]

Which issue does this PR close?

Part of #18217

Rationale for this change

In FilterExec, selectivity is calculated as output_rows/input_rows.
This PR supports such metric. I think this metrics provides important application-level insights, and would be commonly used, so it is displayed in the summary verbose level.

Demo in datafusion-cli

> set datafusion.explain.analyze_level = summary;
0 row(s) fetched.
Elapsed 0.000 seconds.

> explain analyze select * from generate_series(100) as t1(v1) where v1 <10;
+-------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| plan_type         | plan                                                                                                                                                                                   |
+-------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Plan with Metrics | ProjectionExec: expr=[value@0 as v1], metrics=[output_rows=10, elapsed_compute=1.763µs, output_bytes=64.0 KB]                                                                          |
|                   |   CoalesceBatchesExec: target_batch_size=8192, metrics=[output_rows=10, elapsed_compute=25.833µs, output_bytes=64.0 KB]                                                                |
|                   |     FilterExec: value@0 < 10, metrics=[output_rows=10, elapsed_compute=34.888µs, output_bytes=128.0 B, selectivity=9.9% (10/101)]                                                      |
|                   |       RepartitionExec: partitioning=RoundRobinBatch(14), input_partitions=1, metrics=[]                                                                                                |
|                   |         LazyMemoryExec: partitions=1, batch_generators=[generate_series: start=0, end=100, batch_size=8192], metrics=[output_rows=101, elapsed_compute=33.167µs, output_bytes=64.0 KB] |
|                   |                                                                                                                                                                                        |
+-------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row(s) fetched.
Elapsed 0.004 seconds.

What changes are included in this PR?

1. Add a new MetricValue for ratio.
2. Tracking selectivity in FilterExec with MetricValue::Ratio

Are these changes tested?

UT

Are there any user-facing changes?

No

[alamb]

This makes sense to me, though I have a suggestion to reduce the duplication between this and pruned metrics

[alamb]

this is basically the same as Pruned metrics except the display is different -- I wonder if we could consolidate the two somehow 🤔

[2010YOUY01]

Indeed the core logic is mostly duplicated, but I think the tradeoff is more lines of code for better readability, the APIs named add_matched/add_pruned for pruning are easier to understand than the APIs in the ratio metric.

[xudong963]

Thank you, love this one.

[2010YOUY01]

Thank you @alamb and @xudong963 for the review!