Stam

Bugfix release:
* fix for annotations_no_substores() and similar functions

* Breaking API changes:
* `TextResourceBuilder` was simplified, the `build()` method is no longer public
* `AnnotationStore.add()` got renamed to `AnnotationStore.with_item()` but is a lower-level function that should be used less.
* Use the new `AnnotationStore.add_resource()` and `AnnotationStore.add_dataset()` instead (or the variants `with_resource()` and
`with_dataset()`).
* `include` behaviour is now more strictly defined (STAM v1.1.1). Relative files are always interpreted relative to the working directory (either an explicitly set working directory in the configuration, or the actual system's cwd) (https://github.com/annotation/stam/issues/31)
* API changes
* Expanded substore API, added `AnnotationStore.annotations_no_substores()`, `Annotation.substore()`, `TextResource.substores()`, `AnnotationDataSets.substores()`
* Added `AnnotationDataSetBuilder`, to be used with `AnnotationStore.add_dataset()`
* query: Added support for substores as contraints (`SUBSTORE` keyword).
* Fixes for https://github.com/annotation/stam/issues/31
* Fixed relative path handling (`include mechanism`)
* Deserialisation fixes for `include` mechanism.
* Substores can have multiple parents

* Query
* Removed QueryNames, no longer needed
* Allow multiple subqueries (https://github.com/annotation/stam/issues/28)
* `subquery()` is now `subqueries()` (iterator).
* Implemented support for custom attributes on queries and constraints (used e.g. by stam view in stam-tools)
* Allow DATASET as primary constraint for KEY result type
* Implemented OPTIONAL qualifier for SELECT subqueries
* Added `Query.bind_from_result()` method to bind variables from a previous result
* DataValue/DataOperator: fixes in parsing disjunctions
* DataValue/DataOperator: allow matching some distinct types if the reference is a string
* Implemented substores; allows including other (stand-off) annotation stores as dependencies, keeping STAM JSON serialisations apart (https://github.com/annotation/stam/issues/29)

fix in STAM JSON serialisation of resources and annotation data sets after deletions

* Fixes a bug in STAMQL query parser

New features:
* Implemented text validation extension 5
* Support LIMIT keyword in query language (annotation/stam25)

Fixes:
* Preceeds and Succeeds by default allow whitespace in between now (17)

API improvements:
* added methods value() and value_as_str() on iterators over AnnotationData