Support a maximum batch size per DataSource

[peterneyens]

• Add the maxBatchSizePhase interpreter which splits fetches to data sources with a maximum batch size in to multiple sequential fetches. This required a change to Concurrent to return a InMemoryCache instead of a DataSourceCache making it possible to combine multiple caches for sequential batches.
• Refactor the processConcurrent method which is used in the coreInterpreter (previously called interpreter) by using XorT instead of nested if / else and Option.fold and by extracting some helper methods.
• Give some members of the Fetch constructors a better name (eg. FetchMany.as to FetchMany.ids)
• Create the Fetch.multiple method which makes it possible to create a Fetch[List[...]] directly (using FetchMany underneath). This can currently (and previously) be done by using sequence or traverse, but multiple doesn't have the overhead of inspecting the Fetch structure to find independent fetches which can be parallelized.

[codecov-io]

Current coverage is 75.87% (diff: 92.04%)

Merging #76 into master will increase coverage by 6.13%

@@             master        #76   diff @@
==========================================
  Files             8          8          
  Lines           195        228    +33   
  Methods         190        226    +36   
  Messages          0          0          
  Branches          5          2     -3   
==========================================
+ Hits            136        173    +37   
+ Misses           59         55     -4   
  Partials          0          0          

Powered by Codecov. Last update 26f7b45...cb60e82

[ghost]

Outstanding work, well done!