Implement abbreviation system to avoid very-long-line outputs

Based on the idea from #53. The abbreviation system has these main components: - Abbreviation levels mark layers at which we should increment a counter, for "levels of depth" into an object. Usually each foldable tree node should increase the level of depth. - Abbreviatable parts indicate parts of an output that should change if the abbreviation level is too high. - The abbreviation threshold (one for regular and one for roundtrip) specifies the number of abbreviation levels *past* the first collaped tree node where abbreviation should apply. Configurable as follows: ```python # as part of the basic setup treescope.basic_interactive_setup( autovisualize_arrays=True, abbreviation_threshold=1, ) # set globally treescope.abbreviation_threshold.set_globally(1) # override in a context with treescope.abbreviation_threshold.set_scoped(1): print(treescope.render_to_text(nested)) ``` Produces outputs like this in HTML mode: <img width="880" alt="image" src="https://github.com/user-attachments/assets/ace123f6-1365-45a9-8da2-a3f8c108e096" /> or like this in text mode: ```python { 'x0': { 'foo': <np.ndarray float64(100, 100) ≈-0.0084 ±0.99 [≥-3.6, ≤3.8] nonzero:10_000>, 'bar': {'x': 1111, 'y': 2222, 'z': 3333, 'c': Config(<6 fields...>)}, 'baz': [3, 3, 3, 3, 3, 3, 3], }, 'x1': { 'foo': <np.ndarray float64(100, 100) ≈-0.02 ±0.99 [≥-4.2, ≤3.6] nonzero:10_000>, 'bar': {'x': 1111, 'y': 2222, 'z': 3333, 'c': Config(<6 fields...>)}, 'baz': [3, 3, 3, 3, 3, 3, 3], }, 'x2': { 'foo': <np.ndarray float64(100, 100) ≈0.0071 ±1.0 [≥-3.6, ≤3.7] nonzero:10_000>, 'bar': {'x': 1111, 'y': 2222, 'z': 3333, 'c': Config(<6 fields...>)}, 'baz': [3, 3, 3, 3, 3, 3, 3], }, 'x3': {'foo': <np float64(100, 100)>, 'bar': {<4 items...>}, 'baz': [<7 elements...>]}, 'x4': {'foo': <np float64(100, 100)>, 'bar': {<4 items...>}, 'baz': [<7 elements...>]}, } ```