Add element of randomness in words chosen for incdec()

[stanley-rudkin-moj]

Can't take credit for this suggestion, because someone who left our team implemented it in our own RAP. But basically the idea is to get incdec() to randomly choose from a set of phrases each time it is called, so that any text written with incdec() doesn't sound too repetitive.

To ensure reproducibility, I added a new argument, seed, which is a random seed used within the function only. Any existing
seed in your R environment will be restored after the function is called. Its default is NULL, which returns different results each time the function is called. But it can be set to a single number to guarantee it returns the same results every time.

For example, incdec(-1, "past", seed = 14) always returns "has fallen by". But incdec(-1, "past") can return any one of "has fallen by", "is down by", or "decreased by".

I'm open to changing the default for seed to a single number which means the original phrases ( e.g. "increased by") are always used, thereby not affecting any team's existing code. But felt that this change at least shouldn't break any code if not.

[stanley-rudkin-moj]

Could even make similar changes to change_desc() so it doesn't always return "increased/decreased"?

[stanley-rudkin-moj]

And also change_desc() could potentially return the value itself as well as the phrase, e.g. "saw an increase of 3%"