We agreed that, when xyz happens, we would do abc
This sentence does not make sense to me. If ‘happens’ is present tense, xyz clearly is still in the future (or is a general statement, ‘when’ meaning in effect ‘whenever’). Therefore, ‘do’ should also be in the future (or present) tense, rather than in the conditional mood:
We agreed that, when xyz happens, we will do abc
We agreed that, when xyz happens, we do abc
Having ‘happens’ in the present, but ‘do’ in the conditional creates an impossible time line, and the sentence collapses semantically. (Although in colloquial language use, I’m sure you’d be quite likely to hear it used)
However, since you say that xyz was future when the agreement was made, but is now past, this is not the proper option to begin with.
We agreed that, when xyz happened, we would do abc
I’m not sure what contingency it is you feel is not expressed here. If you know for a fact that xyz has in fact happened at the time of reporting, there is no contingency to express: it is an at the time future event that is now marked as having actually happened.
In general, I feel the sentence would flow better if the latter two phrases were reversed:
We agreed that we would do abc when xyz happened
But that is a matter of style alone.
As for your last question, I am not aware of any particular grammatical term for the interaction between such constructions, but I’m sure someone has coined one somewhere.
The short answer to your question is "no" the longer answer is "we're working on it." The rabbit hole here gets very deep but it also a really exciting aspect of modern language. Deciding that one word sounds better than another is something that human beings do millions of times when writing. No one sits down and parses through a thesaurus, analyzes historical context, looks for similar usages and then, based on this process, plucks the perfect word out of a list of 30 that mean the same thing. Yet people somehow make these choices all the time and generally agree that some writing feels 'better' or more 'eloquent' than other writing, without being able to pin down exactly why.
One paper that doesn't do exactly what you are looking for, but shows how it might be done is this study of politeness in language done by Stanford another project by the same department is a program that does the exact opposite of what you are looking for (it converts language to logical axioms). Some commercial software can also gauge the degree to which words are positive or negative or find underlying psychological associations with certain terms. These are largely directed towards marketing analysis and other more profitable fields so they often cost a ton.
Best Answer
The noun and verb for this kind of behaviour is a stampede. Here's the relevant definition for the noun from the Oxford English Dictionary (not ODO):
When used in its literal sense, not its figurative one, a stampede tends to imply the injury or death of some people. However, this is not part of the definition. Most people consider the main cause of death during a stampede to be trampling. Here is the relevant definition of trample from OED:
However, this is in fact a misconception. Most people who die in a stampede die from compressive asphixiation (see under Forces on page 5 here). In ordinary language this basically means being suffocated by being squashed.
Refs:
"stampede, n." OED Online. Oxford University Press, June 2015. Web. 18 July 2015.
"trample, v." OED Online. Oxford University Press, June 2015. Web. 18 July 2015.