fix: support 1d vis table conversion

When some visualizations (e.g.: Headline) get executed,
the resulting data is only 1D (single dimension in resultSpec).

In such cases we want to ask for 1d offsets&limits, even if we
explicitly request "always_two_dimensional" table.

risk: low
JIRA: CQ-1105

Author: Tjev

gooddata-sdk/gooddata_sdk/table.py

@@ -244,7 +244,9 @@ def _as_table(response: ExecutionResponse, always_two_dimensional: bool = False)
     first_page_offset = [0, 0]
     first_page_limit = [_TABLE_ROW_BATCH_SIZE, _MAX_METRICS]
 
-    if not always_two_dimensional:
+    # always adjust paging based on presence of metrics/attrs if not always_two_dimensional
+    # (behavior expected in FDW), otherwise, adjust if response contains only one-dimensional data
+    if not always_two_dimensional or len(response.dimensions) == 1:
         if not response.exec_def.has_attributes():
             # there are no attributes, there shall be at most one row with the metrics, so get that as first page
             first_page_limit = [first_page_limit[1]]