Avoid failing on unsupported filters

From an email conversation with an engineer from Databricks:

(Also, scanning through the source of the connector I think that I can already spot a few places where you'll hit potential trouble in 1.5.x upgrading. For instance, https://github.com/couchbaselabs/couchbase-spark-connector/blob/04a3e1be1119c5042075623b2a2e79357264316c/src/main/scala/com/couchbase/spark/sql/N1QLRelation.scala#L160 is probably going to be an issue if we ever add pushdown support for new filter types; you would be better off just ignoring unsupported filters, since Spark will perform its own layer of filtering on top anyways (just to be safe / to enable you to skip unsupported filters / filters that don't map to your query language))