It looks like the OP is basically using the default dd-wrt iptables & ip6tables rulesets. Since he did not run -vnL the IFs are not shown. That being said the first of the duplicate IPv4 input chain "accept state new" would be for lo the second for br0. As for IPv6 it may be prudent to add lo but honestly it does absolutely nothing. I have 0 packets from lo over IPv6.
For IPv4 tcpmss clamping would be found in the mangle table by default with ddwrt. For IPv6 the module ip6table_mangle is not loaded by default and if I remember correctly was not included in the ddwrt builds until recently (ish). If he were to move tcpmss to mangle he would have to insmod ip6tables_mangle.ko himself. As for it being in the filter forward table I believe it would be kind of difficult to outright state that it is not acceptable when the netfilter documentation itself shows it in the filter forward table. FYI it works just fine where it is by default. http://www.netfilter.org/documentation/HOWTO/netfilter-extensions-HOWTO-4.html