fread support for common big data formats

This is a feature request to implement AVRO and ORC reads into datatable. The ORC read function would be akin to the Pandas implementation:

```python
df = pd.read_orc("my_orc_file")
```

To my knowledge, there is currently no fast way to read an avro file into Pandas. The fastest way I am familiar with is:

```python
import fastavro
with open(file, "rb") as f:
reader = fastavro.reader(f)
records = [r for r in reader]
df = pd.DataFrame.from_records(records)
```