feat: comet native scan improvements - Dynamic Partition Pruning

[Shekharrajak]

Which issue does this PR close?

Ref #3510

Rationale for this change

CometNativeScanExec currently falls back to Spark when Dynamic Partition Pruning (DPP) is present. This limits performance for star-schema queries that rely on DPP to prune partitions at
runtime based on dimension table filters.

What changes are included in this PR?

Added DPP support to CometNativeScanExec for V1 native scans
Implemented partition filter evaluation from DPP subqueries

How are these changes tested?

Added DPP benchmark comparing Spark vs Comet native scan performance
Unit tests

$ make benchmark-org.apache.spark.sql.benchmark.CometDPPBenchmark

  OpenJDK 64-Bit Server VM 17.0.13+11 on Mac OS X 26.2                                                                                                                                       
  Apple M4 Max
  Star-Schema DPP Query (5000000 rows, 50 partitions):  Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative                                                       
  -----------------------------------------------------------------------------------------------------------------------------------
  Spark (JVM) with DPP                                           163            173           9         30.7          32.6       1.0X
  Spark (JVM) without DPP                                        154            164           7         32.4          30.9       1.1X
  Comet (Native) with DPP                                        103            107           3         48.7          20.5       1.6X
  Comet (Native) without DPP                                      98            101           2         50.9          19.6       1.7X

[Shekharrajak]

CI checks where failing and hence need to update them

[andygrove]

I'd suggest not changing the fallback message for this PR and have a follow on PR to improve the message, so that this PR is smaller and just focuses on the functionality.

Another option is to add a new config to feature gate the DPP support and disable it for now in the stability suite.

[Shekharrajak]

Thanks @andygrove for review, Actually unit tests were failing and hence I had to update in PR itself to make all checks green.

[andygrove]

👍

[Shekharrajak]

Please trigger the CI checks

[Shekharrajak]

CI check failure is due to network issue :

 Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
100 43705  100 43705    0     0  2538k      0 --:--:-- --:--:-- --:--:-- 2667k
Downloading https://services.gradle.org/distributions/gradle-8.13-bin.zip

Error: Exception in thread "main" java.io.IOException: Server returned HTTP response code: 502 for URL: https://github.com/gradle/gradle-distributions/releases/download/v8.13.0/gradle-8.13-bin.zip
	at java.base/sun.net.www.protocol.http.HttpURLConnection.getInputStream0(HttpURLConnection.java:1978)
	at java.base/sun.net.www.protocol.http.HttpURLConnection.getInputStream(HttpURLConnection.java:1564)
	at java.base/sun.net.www.protocol.https.HttpsURLConnectionImpl.getInputStream(HttpsURLConnectionImpl.java:250)
	at org.gradle.wrapper.Install.forceFetch(SourceFile:2)
	at org.gradle.wrapper.Install$1.call(SourceFile:8)
	at org.gradle.wrapper.GradleWrapperMain.main(SourceFile:67)
Error: Process completed with exit code 1.

How can we re-trigger those 2 failing CI workflow ?