here are some improvement about TSDB index.routing_path settings:

1. support custom routing search:

• if user set only one field routing_path, it can search with: GET xxx/_search?routing=xx
• if user set multi fields in routing_path, it can search with: GET xxx/_search?routing=filed1:xxx|field2:xxx|field3:xxx
• if user set wildcard fields in routing_path, it can reject the routing parameter

2.ignore object type check:

• It will cause that some mistake. e.g: {"foo":{"bar":"xxx"}} , index.routing_path : "foo.*" is ok, but index.routing_path : "foo*" is not ok.
• I think we can ignore object type, because the routing_path will check the field if it has time_series_dimension parameter, as object can't set the parameter, it can ignore the object check.

3.how to deal with ''.":

• e.g index.routing_path : "foo.bar", the document is ok:

{
  "foo": {
    "bar": "xxx"
  }
}

but the document is failed:

{
  "foo.bar" : "xxx"
}

This can be confusing for the user, because the user configures index.routing_path : "foo.*", but the field foo.bar is failed, user must set index.routing_path : "foo\.*" to support field foo.bar.

The reason why foo.bar is failed, is because of the Xcontent filter rule. May be it can be improved.