Update IMU noise params for simulation

[2shaar2059]

I ran an Allan Variance analysis on the BNO085 gyro data. @perrintong325 collected the data.

Allan Variance Analysis output:

====================================================================================================
gyro_x
Estimated Bias: 705.698536811248 [deg/hr]
Estimated Angle Random walk:31.275580469753645 [(deg/hr) / sqrt(Hz)]
         <==> 978.1619337200357 [(deg/hr)^2 / Hz] or [(deg/hr)^2 * s]
         <==> 0.2717116482555655 [deg^2/hr]
         <==> 0.521259674495894 [deg/sqrt(hr)]
Estimated Angular Rate Random walk:2.0809041100619394 [((deg/hr)/s) / sqrt(Hz)]
         <==> 4.330161915272672 [((deg/hr)/s)^2/Hz] or [((deg/hr)/s)^2 * s] or [(deg/hr)^2 / s]
         <==> 15588.58289498162 [(deg/hr)^2 / hr]
         <==> 124.85424660371636 [(deg/hr) / sqrt(hr)]
====================================================================================================
gyro_y
Estimated Bias: -531.5969709753438 [deg/hr]
Estimated Angle Random walk:25.023971267068088 [(deg/hr) / sqrt(Hz)]
         <==> 626.1991379750492 [(deg/hr)^2 / Hz] or [(deg/hr)^2 * s]
         <==> 0.17394420499306923 [deg^2/hr]
         <==> 0.41706618778446813 [deg/sqrt(hr)]
Estimated Angular Rate Random walk:0.3242245968154019 [((deg/hr)/s) / sqrt(Hz)]
         <==> 0.10512158918010991 [((deg/hr)/s)^2/Hz] or [((deg/hr)/s)^2 * s] or [(deg/hr)^2 / s]
         <==> 378.4377210483957 [(deg/hr)^2 / hr]
         <==> 19.453475808924114 [(deg/hr) / sqrt(hr)]
====================================================================================================
gyro_z
Estimated Bias: 45.57008661175387 [deg/hr]
Estimated Angle Random walk:32.232438622600405 [(deg/hr) / sqrt(Hz)]
         <==> 1038.9300995597023 [(deg/hr)^2 / Hz] or [(deg/hr)^2 * s]
         <==> 0.28859169432213955 [deg^2/hr]
         <==> 0.5372073103766735 [deg/sqrt(hr)]
Estimated Angular Rate Random walk:0.4422940047256151 [((deg/hr)/s) / sqrt(Hz)]
         <==> 0.19562398661622243 [((deg/hr)/s)^2/Hz] or [((deg/hr)/s)^2 * s] or [(deg/hr)^2 / s]
         <==> 704.2463518184007 [(deg/hr)^2 / hr]
         <==> 26.537640283536906 [(deg/hr) / sqrt(hr)]
====================================================================================================

Measured Noise Ranges:

initial bias: -531 to 706 [deg/hr]
angle random walk: 0.4 to 0.5 [deg/sqrt(hr)]
rate random walk (angle random run): 19 to 125 [(deg/hr) / sqrt(hr)]

Suggested Noise dispersions in montecarlo:

initial_bias_range = np.deg2rad([-1000, 1000])/3600  # [rad/s]
sigma_v_range = np.deg2rad([0.5 / np.sqrt(60), 2.0 / np.sqrt(60)])  # [rad/sqrt(s)]
sigma_w_range = np.deg2rad([19 / np.sqrt(60), 500 / np.sqrt(60)])/3600  # [(rad/s)/sqrt(s))]

The consequential update is to sigma_w_range. Currently, params.yaml is too pessimistic/conservative on the minimum value for this parameter. The gyro bias was measured to be more stable.