CGMeshExtended: added method adaptiveSubdivide() to refine the mesh depending on the local edge length

added signals VISION_SCENE_UPDATE_CYCLIC_START, VISION_SCENE_UPDATE_CYCLIC_STOP, and VISION_SCENE_UPDATE_SINGLE to give more control on when scene updates are sent from the vision system to the control system

added special output methods for VisionToControlContainer (combining terrain patches and object surfaces and volumes)

TOMLBaseContext: removed assert(false) for handleError() since this aborts a debug build in case of a non-critical input error (e.g. if a table does not exist)

TerrainSpecification: height map and confidence map now use DefaultInitializationAllocator from broccoli (resize now does NOT initialize cells to 0 - increses performance for serialization and encoding)

TOMLDecodingContext: added decodeOptionalValue() and getSubSections() + fixed exception handling in case a value does not exits

splitted TOMLData into TOMLInputData and TOMLOutputData and TOMLFile into TOMLInputFile and TOMLOutputFile

Minifix: removed validity checks for ObjectSurfaceModel (triangle mesh), ObjectVolumeModel (ssv segment) and TerrainPatchUpdate (map dimensions) during serialization (binary) and encoding (TOML) to improve real-time performance

ObjectVolumeModel: fixed dummy SSVTriangleElement construction (now uses valid parameters to avoid assertion failure in DEBUG mode)

changed pose representation of ViconObject from std::array to Eigen::Vector/Eigen::Quaternion (for consistency with the order of coefficients x/y/z and w/x/y/z)

added signals to start/stop scene reconstruction + WIP: started implementation of TOML encoding/decoding

updated API for writing interface containers to binary files + enabled compression for serialization by default

added high-level signals (clearing scene) + added terrain specification and interface objects (environment model)

extended interface from control to vision (added pose data of vision-tcp and torso-imu frame + raw IMU data of torso-mounted IMU)