ENH: Add NGARCH specification to set of volatility processes

[itsNH98]

Actually, the "volatility" component of arch includes numerous GARCH specifications, many of them that include the leverage effect. However, it doesn't include the NGARCH of Engle and Ng (1993).

ARCH does include a lot of different models so this addition could seem unnecessary, but it would mostly be for completeness and replication of prior results.

Thanks

[atharvajoshi01]

Is this still on the roadmap? I'd be happy to implement the NGARCH(1,1) specification following Engle and Ng (1993). The variance equation would be:

h_t = omega + alpha * (r_{t-1} - theta * sqrt(h_{t-1}))^2 + beta * h_{t-1}

where theta captures the leverage effect. Would slot in alongside the existing EGARCH and GJR-GARCH implementations in the volatility module.