Update README to reflect project vision

[andygrove]

This PR updates the README to better explain the features that we plan on supporting:

Rendered version: https://github.com/andygrove/datafusion-ray/blob/new-readme/README.md

[robtandy]

Thank you @andygrove !

A few suggestions:

• Can we call Greedy, Streaming? I think its more communicative about how it functions and conveys its major feature in the name.

• for the code snippet, I think it should read:

  import ray
  from datafusion_ray import DFRayContext
  
  ray.init()
  session = DFRayContext()
  df = session.sql("SELECT * FROM my_table WHERE value > 100")
  df.show()

• I'm not sure about the trade offs as written. I think its possible that, depending on the query, a batch mode could be faster than a streaming mode for smaller queries due to less overhead. We'll have to implement the batch mode to define this more clearly.

• We should indicate that the batch mode is planned for 0.2.0 and 0.1.0 will include Streaming only

[robtandy]

For the code snippet, i forgot to include ray.init(runtime_env=df_ray_runtime_env)

We should have, I think,

import ray
from datafusion_ray import DFRayContext, df_ray_runtime_env

ray.init(runtime_env=df_ray_runtime_env)
session = DFRayContext()
df = session.sql("SELECT * FROM my_table WHERE value > 100")
df.show()

As df_ray_runtime_env is necessary to set up logging correctly in Ray workers

[andygrove]

Thank you @andygrove !

A few suggestions:

* Can we call `Greedy`, `Streaming`?   I think its more communicative about how it functions and conveys its major feature in the name.

* for the code snippet, I think it should read:
  ```python
  import ray
  from datafusion_ray import DFRayContext
  
  ray.init()
  session = DFRayContext()
  df = session.sql("SELECT * FROM my_table WHERE value > 100")
  df.show()
  ```

* I'm not sure about the trade offs as written.   I think its possible that, depending on the query,  a batch mode could be faster than a streaming mode for smaller queries due to less overhead.  We'll have to implement the batch mode to define this more clearly.

* We should indicate that the batch mode is planned for `0.2.0` and `0.1.0` will include Streaming only

I made the following updates:

• Use Streaming and Batch terminology
• Added note that batch is not implemented yet, with link to the tracking issue
• Updated code example

Do you have specific suggestions for updating the trade-offs?