Parking location bug fix

[i-am-sijia]

This PR is to address the bug in parking location choice documented in #620.

The bug fix was done as part of the MTC TM2 ActivitySim implementation. It will be reviewed and adopted by SANDAG ActivitySim implementation. @kulshresthaa made the original code changes and his commit history can be found here.

[jpn--]

So, this code change appears consistent with the intent and from my review of the results in the linked BayDAG repo, it appears to fix the problem.

But: this PR changes the parking location choice model, without touching a single thing in the ARC example implementation -- no config.yaml changes, no spec.csv changes, and, most of all, no changes to any ARC test setup or regression test outcome. And yet, all the tests still pass. This suggests to me that (a) this code change is effectively not being tested adequately in this new PR, and (b) the parking location choice model is probably not being adequately exercised by any test prior to this PR either.

[i-am-sijia]

So, this code change appears consistent with the intent and from my review of the results in the linked BayDAG repo, it appears to fix the problem.

But: this PR changes the parking location choice model, without touching a single thing in the ARC example implementation -- no config.yaml changes, no spec.csv changes, and, most of all, no changes to any ARC test setup or regression test outcome. And yet, all the tests still pass. This suggests to me that (a) this code change is effectively not being tested adequately in this new PR, and (b) the parking location choice model is probably not being adequately exercised by any test prior to this PR either.

We did not need to update the parking location UEC for the actual ARC implementation, the code change fixed the bug. The parking location results after this update matches ARC's CTRAMP implementation, as Raghu noted in his follow-up comments in #620.

I took a look at the prototype ARC trip list, there are no trips having actual parking zone values - they all park at the destination zones - both before and after this change, which explains why the tests still pass. To your point, parking location choice model is not being adequately tested with the current prototype model.