Split shared memory layout into metric/non-metric part #4873
Description
Current our shared memory is a singled block of annotation data and non-metric data. We should split this into one block responsible for persistent data (e.g. names, graph structure, axillary data) and data depending on the metric being used.
Splitting this data into two updatable pieces is the first step towards #4007.
Metric independent:
"EDGE_BASED_NODE_DATA",
"ANNOTATION_DATA",
"COORDINATE_LIST",
"OSM_NODE_ID_LIST",
"TURN_INSTRUCTION",
"ENTRY_CLASSID",
"R_SEARCH_TREE",
"R_SEARCH_TREE_LEVELS",
"GEOMETRIES_INDEX",
"GEOMETRIES_NODE_LIST",
"GEOMETRIES_FWD_DURATION_LIST",
"GEOMETRIES_REV_DURATION_LIST",
"GEOMETRIES_FWD_DATASOURCES_LIST",
"GEOMETRIES_REV_DATASOURCES_LIST",
"HSGR_CHECKSUM",
"TIMESTAMP",
"FILE_INDEX_PATH",
"DATASOURCES_NAMES",
"PROPERTIES",
"BEARING_CLASSID",
"BEARING_OFFSETS",
"BEARING_BLOCKS",
"BEARING_VALUES",
"ENTRY_CLASS",
"LANE_DATA_ID",
"PRE_TURN_BEARING",
"POST_TURN_BEARING",
"TURN_LANE_DATA",
"LANE_DESCRIPTION_OFFSETS",
"LANE_DESCRIPTION_MASKS",
"TURN_DURATION_PENALTIES",
"MLD_LEVEL_DATA",
"MLD_PARTITION",
"MLD_CELL_TO_CHILDREN",
"MLD_CELL_SOURCE_BOUNDARY",
"MLD_CELL_DESTINATION_BOUNDARY",
"MLD_CELLS",
"MLD_CELL_LEVEL_OFFSETS",
"MLD_GRAPH_NODE_LIST",
"MLD_GRAPH_EDGE_LIST",
"MLD_GRAPH_NODE_TO_OFFSET",
"MANEUVER_OVERRIDES",
"MANEUVER_OVERRIDE_NODE_SEQUENCES"
Metric dependent:
"CH_GRAPH_NODE_LIST",
"CH_GRAPH_EDGE_LIST",
"CH_EDGE_FILTER_0",
"CH_EDGE_FILTER_1",
"CH_EDGE_FILTER_2",
"CH_EDGE_FILTER_3",
"CH_EDGE_FILTER_4",
"CH_EDGE_FILTER_5",
"CH_EDGE_FILTER_6",
"CH_EDGE_FILTER_7",
"GEOMETRIES_FWD_WEIGHT_LIST",
"GEOMETRIES_REV_WEIGHT_LIST",
"TURN_WEIGHT_PENALTIES",
"MLD_CELL_WEIGHTS_0",
"MLD_CELL_WEIGHTS_1",
"MLD_CELL_WEIGHTS_2",
"MLD_CELL_WEIGHTS_3",
"MLD_CELL_WEIGHTS_4",
"MLD_CELL_WEIGHTS_5",
"MLD_CELL_WEIGHTS_6",
"MLD_CELL_WEIGHTS_7",
"MLD_CELL_DURATIONS_0",
"MLD_CELL_DURATIONS_1",
"MLD_CELL_DURATIONS_2",
"MLD_CELL_DURATIONS_3",
"MLD_CELL_DURATIONS_4",
"MLD_CELL_DURATIONS_5",
"MLD_CELL_DURATIONS_6",
"MLD_CELL_DURATIONS_7",