Fix stringop-overread compile issues

[svwilliams]

When compiling ROS2 Rolling on Ubuntu Noble 24.04 using GCC 13.2.0, there are many instances of the following compiler error:

/usr/lib/gcc/x86_64-linux-gnu/13/include/emmintrin.h:169:19: error: ‘void* __builtin_memcpy(void*, const void*, long unsigned int)’ reading 16 or more bytes from a region of size 8 [-Werror=stringop-overread]
  169 |   *(__m128d *)__P = __A;

This seems similar to a previously encountered compiler bug in GCC 12

# include/avxintrin.h:893:24: error: array subscript ‘__m256d_u[0]’ is partly outside array bounds of ‘Eigen::internal::plain_matrix_type<Eigen::Product<Eigen::Matrix<double, 3, 3, 1>, Eigen::Map<Eigen::Matrix<double, -1, 1> >, 0>, Eigen::Dense>::type [1]’ {aka ‘Eigen::Matrix<double, 3, 1> [1]’} [-Werror=array-bounds]
#   893 |   return *(__m256d_u *)__P;

But I do not see any open bug tickets about this issue in either Eigen or GCC. This issue may be entirely harmless; I'm really not sure.

However, every instance seems to occur when an Eigen Fixed-sized object is being assigned to an Eigen Dynamic-sized object. E.g. some code along the lines of:

fuse_core::Vector3d a;
a << 1, 2, 3;
fuse_core::VectorXd b = a;

A vast majority of these assignments occur in unit tests, so fixing those is trivial. I have no concerns about runtime at all:

fuse_core::VectorXd a(3);
a << 1, 2, 3;
fuse_core::VectorXd b = a;

There are a couple of places in actual library code where I update what type of Eigen object is being created, in sensor_proc.hpp and unicycle_2d_ignition.cpp. Both of those files are in the fuse_models package. I do not expect a meaningful performance impact from either of these changes.

And one instance in normal_delta_orientation_3d_cost_functor.hpp in the fuse_constraints package. That one should probably be classified as a bug. The size is known at compile-time; the Eigen::Map should never have be configured for a dynamic size.

Also, the CI build tests are going to fail. There is an API change to the rclcpp::Waitable class that is being fixed here: #379

[DavidLocus]

LGTM

[henrygerardmoore]

Applying this on top of #379 allowed me to build in release and debug, and fixed the stringop-overread compile problems. 👍