There is a way to manage tables in rules, but there is no way to
manage tables for chains when setting policy.

Looks like pillar structure is bad from the beginning and to
not break backward compatibility, as same chain names may occur in
different tables, so it is proposed to check if 'chain.policy' is
map. And if it is, specific policies would be ensured for specific
tables, otherwise table 'filter' would be used as a fallback.

To ensure chains in specific tables we iterate over all rules in
each chain.

This hash is valid:

parameters:
  iptables:
    service:
      enabled: true
      chain:
        OUTPUT:
          policy: ACCEPT
        FORWARD:
          policy:
          - table: mangle
            policy: DROP
        INPUT:
          policy:
          - table: nat
            policy: ACCEPT
          rules:
          - jump: ACCEPT
            protocol: icmp
        POSTROUTING:
          rules:
          - jump: MASQUERADE
            protocol: icmp
            out_interface: ens3
            table: nat

Prod-Related: CEEMCP-12
Prod-Related: EME-313

Change-Id: Ib5ba97dad165d3ef2dec7e053b391ea36a996103

* change file mode to 644

* change file mode to 644

Change-Id: I9d93052cfc197a364b42240448344d5543e8805f

Closes: #2 

* add the option to specify the family per rule to support ipv6

* include policy updates for ipv6

* update documentation to mention ipv6

* Make ipv6 optional; remove spurious tabs from the readme.

* set ipv6 policies only if ipv6 is enabled on the host and not explicitly turned off for this service