Refactor CIDRMATCH function for SQL/PPL to use single implementation

**Is your feature request related to a problem?**
Currently, CIDRMATCH function has separate implementation for SQL and PPL
Refer: https://github.com/opensearch-project/opensearch-spark/pull/1100

**What solution would you like?**
Refactor to have single shared implementation.

**What alternatives have you considered?**
n/a

**Do you have any additional context?**
n/a