Unknown Flavor, excaliber json incompatible with camelot

**Describe the bug**
When using the Excalibur to create JSON rules, it is impossible to use them in the **kwargs of camlot cli.

**Steps to reproduce the bug**
1. Upload a file to Excalibur
2 Generate extraction rules in excalibur
3 Go to the Excalibur Rule manager. Download or copy the JSON rule
Example:

```
 {"pages":{"1":{"table_areas":["83.39951690821256,619.54177384945,509.9832528180354,466.19127305596726"],"columns":null}},"flavor":"Stream","process_background":false,"line_size_scaling":15,"split_text":false,"flag_size":false}
```

4 Call camelot fro the cli with the previously generated JSON rule as kwargs.

This results in:
```
camelot.read_pdf(invoicefile, **kwargs)
  File "/home/user/.local/lib/python3.7/site-packages/camelot/io.py", line 103, in read_pdf
    "Unknown flavor specified." " Use either 'lattice' or 'stream'"
NotImplementedError: Unknown flavor specified. Use either 'lattice' or 'stream'
```

Seems that Excalibur outputs Lattice and Stream.
(Note the uppercase) Camelot CLI expects the flavors to be in lowercase.

So the workaround is simple. Change the Capitol letter of flavor name in the Excaliber JSON to lowercase.
It's just not so good partice to do in real life workflow.

**Expected behavior**
Expected behaviour is that one could use Excaliber to generate rules.
Use an external program in your ETL pipline to call camelot CLI with the setting generator with Excalibur.

**Code**
https://github.com/camelot-dev/camelot/blob/master/camelot/io.py    line 101

```
import camelot

# add your code here
```

**PDF**

**Screenshots**

**Environment**

- OS: ubuntu
- Python version: 3.7
- Camelot version: 0.10.1
- Excalibur version: ??

**Additional context**

Start at camelot/io.py around line 101, where the flavor value from kwargs is validated, and compare it with the uppercase flavor values produced by Excalibur's JSON rules. Reproduce the failure using the example rule and confirm that a rule generated by Excalibur can be passed to camelot.read_pdf without manually editing the flavor value.

Written by the indexing model from the issue text.