add improved component container for image pipeline

[gezp]

解决https://github.com/robomaster-oss/rmoss_core/issues/8.

增加基于CamServerManager的component container，实现容器内节点之间图像直接传输，而不经过ROS通信机制。

• 定义新的CamNodeFactory (继承rclcpp_components::NodeFactory), 定义新的的API，将CamServerManager传递给Node节点
• 定义新的CamComponentManager (继承rclcpp_components::ComponentManager)，覆盖create_component_factory() API, ，优先尝试加载CamNodeFactory类型，否则加载普通rclcpp_components类型节点。

[gezp]

benchmark test for image pipeline

• 采用图像模拟相机节点，发布频率为100hz.
• 测试平均延迟：从CamServer图像采集完成到CamClient接收到图像的时间（采用ROS topic方式，也就是ROS传输数据时间）

测试命令：

# 下载代码
git clone https://github.com/robomaster-oss/rmoss_interfaces.git
git clone -b rmoss_cam_component_container  https://github.com/robomaster-oss/rmoss_core.git 
# 编译后运行命令
ros2 launch rmoss_cam benchmark.launch.py 

限于ROS Galactic版本的rclcpp_compoenet 并没有合并Rolling特性（为每一个Node分配一个独立专用线程），因此采用manual composition进行测试。在Intel(R) i7-8700主机上测试结果如下：

CPU: Intel(R) i7-8700 (6Cores 12Threads), Memory: 32GB	delay time(s)
normal process	0.00249732
manual composition	0.00120593
manual composition with CamServerManager	0.000213741

使用CamServerManager 的manual composition的延迟比多进程ROS节点方法降低了90%左右，具有明显性能提升。

[TomwKang]

设备信息 NUC8 i5 bek CPU： Intel® Core™ i5-8259U CPU @ 2.30GHz × 8 Mem： 7.6 GiB

[benchmark_test-2] start test for normal process
[benchmark_test-2] result for <benchmark_cam1>:
[benchmark_test-2] fps:87.4122
[benchmark_test-2] delay:0.00527591
[benchmark_test-2] start test for normal component
[benchmark_test-2] result for <benchmark_cam2>:
[benchmark_test-2] fps:100.224
[benchmark_test-2] delay:0.00173974
[benchmark_test-2] start test for intra-component
[benchmark_test-2] result for <benchmark_cam3>:
[benchmark_test-2] fps:100.181
[benchmark_test-2] delay:0.00064917

[benchmark_test-2] start test for normal process
[benchmark_test-2] result for <benchmark_cam1>:
[benchmark_test-2] fps:92.7605
[benchmark_test-2] delay:0.00308627
[benchmark_test-2] start test for normal component
[benchmark_test-2] result for <benchmark_cam2>:
[benchmark_test-2] fps:100.195
[benchmark_test-2] delay:0.00188324
[benchmark_test-2] start test for intra-component
[benchmark_test-2] result for <benchmark_cam3>:
[benchmark_test-2] fps:100.198
[benchmark_test-2] delay:0.000420136

[benchmark_test-2] start test for normal process
[benchmark_test-2] result for <benchmark_cam1>:
[benchmark_test-2] fps:99.6199
[benchmark_test-2] delay:0.00496471
[benchmark_test-2] start test for normal component
[benchmark_test-2] result for <benchmark_cam2>:
[benchmark_test-2] fps:100.202
[benchmark_test-2] delay:0.00405021
[benchmark_test-2] start test for intra-component
[benchmark_test-2] result for <benchmark_cam3>:
[benchmark_test-2] fps:100.194
[benchmark_test-2] delay:0.00104779

[gezp]

test result on Raspberrypi 4B with 2GB RAM (ARM platform)

[benchmark_test-2] start test for normal process
[benchmark_test-2] result for <benchmark_cam1>:
[benchmark_test-2] fps:33.8542
[benchmark_test-2] delay:0.0388991
[benchmark_test-2] start test for normal component
[benchmark_test-2] result for <benchmark_cam2>:
[benchmark_test-2] fps:73.3764
[benchmark_test-2] delay:0.0212094
[benchmark_test-2] start test for intra-component
[benchmark_test-2] result for <benchmark_cam3>:
[benchmark_test-2] fps:100.203
[benchmark_test-2] delay:0.002392