KSQL should support STRUCT fields for TIMESTAMP

KSQL only supports top-level fields in the `WITH (TIMESTAMP` clause when defining a stream against an existing topic. It should also support nested fields.

For example, given source message

{"type":"key1","data":{"ts":"2018-11-20 19:20:21.1","a":1,"b":"hello"}}

it should be possible to execute:

```
CREATE STREAM T2 (TYPE VARCHAR, \
DATA STRUCT< \
TS VARCHAR, \
A INT, \
B VARCHAR, \
C INT, \
D VARCHAR>) \
WITH (KAFKA_TOPIC='t_raw',\
VALUE_FORMAT='JSON', TIMESTAMP='DATA->TS', TIMESTAMP_FORMAT='yyyy-MM-dd HH:mm:ss.S');
```

but it fails:

No column with the provided timestamp column name in the WITH clause, DATA->TS, exists in the defined schema.

Start by reproducing the CREATE STREAM example with the nested DATA->TS timestamp expression and inspect how the TIMESTAMP clause resolves fields in the defined schema. Trace the validation from the CREATE STREAM entry point; done means the nested timestamp field is accepted and the example executes without the reported missing-column error.

Written by the indexing model from the issue text.