Add GNC threshold parameter

[doggydoggy0101]

This PR adds the ability to change $\bar{c}$ in the TLS/GM function.

$$ \rho_\text{TLS}(r, \bar{c})=\min(r^2, \bar{c}^2)\ ;\ \ \rho_\text{GM}(r, \bar{c})=\frac{\bar{c}^2r^2}{\bar{c}^2+r^2} $$

---

Originally, in gtsam/nonlinear/GncOptimizer.h, $\bar{c}$ is always set to 1.

void setInlierCostThresholdsAtProbability(const double alpha) {
  barcSq_  = Vector::Ones(nfg_.size()); // initialize <- THIS LINE
  for (size_t k = 0; k < nfg_.size(); k++) {
    if (nfg_[k]) {
      barcSq_[k] = 0.5 * Chi2inv(alpha, nfg_[k]->dim()); // 0.5 derives from the error definition in gtsam
    }
  }
}

We add a barc parameter to gtsam/nonlinear/GncParams.h, thereby we can set barcSq with respect to barc

void setInlierCostThresholdsAtProbability(const double alpha) {
  barcSq_  = (params_.barc * params_.barc) * Vector::Ones(nfg_.size()); // initialize threshold to $bar{c}^2$
  for (size_t k = 0; k < nfg_.size(); k++) {
    if (nfg_[k]) {
      barcSq_[k] = 0.5 * Chi2inv(alpha, nfg_[k]->dim()); // 0.5 derives from the error definition in gtsam
    }
  }
}

We also tested it in examples/GNCExample.cpp

// Set GNC-specific options
GncParams<LevenbergMarquardtParams> gncParams(lmParams);
gncParams.setLossType(GncLossType::TLS);  // Truncated Least Squares or Geman-McClure 
gncParams.setBarc(1.0);                   // Parameter $c$ in Truncated Least Squares or Geman-McClure

---

In addition, there are other ways to set barcSq in gtsam/nonlinear/GncOptimizer.h

void setInlierCostThresholds(const double inth) {
  barcSq_ = inth * Vector::Ones(nfg_.size());
}

void setInlierCostThresholds(const Vector& inthVec) {
  barcSq_ = inthVec;
}

These two functions are not used in the origin GNC implementation, thus I am not sure what inth is. It seems like inth is $\bar{c}^2$ in gtsam/test/testGncOptimizer.cpp and is meant to set to $\bar{c}^2$ directly if known, so I left it unchanged.