Put back the indices measures on create_parameter_hierarchy_from_members()

[alexis-bvq]

Description

Hello,
I am using a parameter hierarchy to choose differente quantile percentages rules.
My first member is called "Decreasing" and will mean that my quantile will follow this rule:
quantile_percentage = 100 - 0.01 * (number_of_years_in_the_future)
Else, members are like "99%", "99.5" and are static of the the years in the future.
It was very handy to have the ability to edit the index measure because this allowed me to put :

indices =     [0, 999, 998, 990,  985, 980,...]
members = ["Decreasing", "99.9%", "99.8%", "99%", "98.5%", "98%", ...]
Therefore I had a measure for the quantile percentage described like this:
m[quantile_percentage] = where(
    l[quantile] == "Decreasing",
    1 - 0.01 * m[time],
    m[indices] / 100.0
)

Because I won't be able to mix string and float within my members it will be way trickier to create the same behaviour in 0.6.2 .

Context

Other information (if relevant)

[alexis-bvq]

Hello, where are we on this requirement ? I face another issue again for that problem.

[alexis-bvq]

other use case being:
you have a sensi vector:
[ 10, 20, 30 ]
To each value is link a time bucket, let's say:
[ 1Y, 2Y, 4Y ] and its number of days [ 365, 730, 1095 ]
It is doable to extend the sensi vector to their bucket.
Nevertheless, the client is taking a step further by bucketing on time that are not in the time bucket and interpolate the value link to those new time buckets.
For example, extended time buckets can be:
[ 1Y, 1.5Y, 2Y, 3.5Y, 4Y]
and therefore the value for each member will be:
[ 10, 15, 20, 27.5, 30 ]
It could be easily done by mapping the number of month/days as a value to each time bucket extended in the parameter hiearchy created before, but we can not anymore.
As you notice the bucket are non linear therefore having only incremental indexes can not work.

[alexis-bvq]

One way of avoiding doing that is by doing a copper join with the mapped elements. Nevetherless you need at least one key in order to do a join and sometimes in the dataset you don't have members with small cardinalities.