[iceberg] Document configuration flags needed for Comet Iceberg to work correctly

[parthchandra]

We need to set -

"spark.comet.use.lazyMaterialization" -> "false"
"spark.comet.schemaEvolution.enabled" -> "true"
"spark.sql.iceberg.parquet.reader-type" -> "COMET"

These need to be documented. Eventually we need to eliminate the need for the user to set these explicitly.

[hsiang-c]

One more in 0.9.1 and 0.10.0 release

"spark.comet.exec.shuffle.enabled" -> "false"

[parthchandra]

Docs updated in #2166