CODE:
* Minor optimizations:
When collecting files to parse we use a hashed set now instead of a vector.
In theory that is way more efficient, in practice it will probably only make a
difference with thousands of files.
Instead of storing the file name in the entity to establish the backward
reference, store the file names in a set and just store a cursor to the
appropriate position in the set. We will need that kind of cross reference
anyway.
* Read steps value for provers and take them into account when sorting. Some
items are now ordered differently, so templates have been updated.
Sidenote: I am not sure how ideal it is to compare steps between different
provers.
* Sorting stability improvement with assigning unique id to sorted items.
* Added steps to output information.
* Fixed inconsistency, the File component had a type Subject_Name instead of
File_Name.
* Moved Image function (instance for Duration) into SPAT package. The useful
types can now all output themselves in the expected format.
* Renamed File/Line/Column functions to Source_* to make intent clear.
DOCUMENTATION:
* Visualization of inheritance tree.
* Script for and final rendering of internal data format.
* + Added 'Saatana' proof output from GNAT CE 2020 for further tests and development.
* * (GNAT CE 2020) Preliminary fix to support summary output.
* + Added version data type and TODO where this should be detected.
* * Implemented crude version detection.
* * Full boolean evaluation for JSON field checks.
* * Support for 'trivial_true' fields and the 'Trivial' prover.
* * Updated documentation for GNAT CE 2020
Marked known differences (fields added, removed).
