[BUG] Coalescing reading is not working for v2 parquet/orc datasource

[wbo4958]

Describe the bug

Even if the input files are not cloud-typed files, or spark.rapids.sql.format.orc/parquet.reader.type is set to COALECING, we still can't get coalescing reading. Instead, multithreaded reading will be used.

The root cause is the default value of queryUsesInputFile which is always true. so the condition canUseCoalesceFilesReader will always be false.

override val canUseCoalesceFilesReader: Boolean =
    rapidsConf.isParquetCoalesceFileReadEnabled && !(queryUsesInputFile || ignoreCorruptFiles)

Steps/Code to reproduce bug

1. read local files
2. check the log, the expected log should be Using the coalesce multi-file. But in fact, the log is Using the multi-threaded multi-file