Add scenarios around pattern matching and self-loops

[Mats-SX]

There is lots of confusion around how self-relationships (or loops) are
matched by patterns. These scenarios aim to outline how this is supposed
to work, and how it differs from simple relationship matching.

[boggle]

I think this does not reflect our last discussion on this, does it? Perhaps something worth picking up w Nadime, esp. MATCH p=(a)-(a) should match each single self-loop exactly once.

[Mats-SX]

@boggle You're right, it doesn't. I'll update it.

[Mats-SX]

Updated with latest thinking. Please also see #192.

[thobe]

Most of these scenarios are fine. But the ones where you could argue that direction would matter are missing. These are scenarios matching paths over loops.

Scenarios like this:

Given an empty graph
And having executed:
  """
  CREATE (a)-[:X]->(a)
  """
When executing query:
  """
  MATCH p=()-->()
  MATCH q=()<--()
  RETURN p = q
  """
Then the result should be:
  | p = q |
  |  ???  |

[thobe]

While I agree with the pattern matching semantics for loops outlined here, I'm not convinced about how this extends to paths in #192, perhaps pattern matching should be considered a separate thing from path matching (at which point perhaps regular path patterns should be required for path matching).