feat: merton's martingale for drift

[rhandal-pfn]

Current drift correction is an approximate of exp(mean + 0.5*sigma^2)-1, which is not an issue for realistic values of jump mean and volatility. The exponential method ensures martingale property for large mu and volatility of the jump.

Changes:

• Use of exponential in jump correction factor
• Use of mean of stock instead of log stock prices to check for martingale property
• Reduce jump amplitude to ensure convergence with small sample path numbers, with a larger jump mean, the number of sample paths also needs to be increased to ensure martingale property.

[rhandal-pfn]

Jump test with mean = 0.1 changes

• changed standard deviation of tolerance based on CLT for mean of paths
• Reduced number of jumps to default jumps per year