Add native step_size support to RangeParameter (#5213)

[saitcakmak]

Summary:

Adds a step_size arg to RangeParameter that snaps values to a grid anchored at lower (in cast()), for both FLOAT and INT parameters. This is the first diff in the step_size unification stack: step_size will subsume both the discrete-grid and limited-resolution (digits) use cases under one knob.

• Next diff will add storage support. The internal DB has already been updated to include the new column.
• We will then migrate all current usage off digits and onto step_size.
• We will add support for treating low-cardinality float-range parameters as discrete in Adapter, so that it is efficiently optimized over the correct grid (rather than having to use continuous optimization + rounding).
• At this point, we will have proper support for step_size, so we can update the ax/api usage to leverage it, rather than resolving to ChoiceParameter.
• We can then deprecate digits and do any remaining clean-up.

In this diff step_size coexists with the existing digits arg (they are mutually exclusive at construction). Subsequent diffs in the stack migrate storage (JSON + SQA), transforms and utils, and the public API (RangeParameterConfig) to step_size, then deprecate digits in favor of it.

Behavior:

• cast() rounds (value - lower) / step_size to the nearest integer and returns lower + n * step_size. It does NOT clamp to [lower, upper]: an out-of-bounds input (e.g. a historical observation recorded outside the current bounds) snaps to the nearest grid point, which may itself be out of bounds. This mirrors the non-step_size cast(), which leaves out-of-bounds values in place rather than silently moving them into range — range validity is enforced by validate(), not cast().
• Both bounds must lie on the grid: (upper - lower) must be an integer multiple of step_size (within EPS). Off-grid bounds are rejected at construction. This guarantees upper is itself a feasible value, so a value near the upper bound snaps to upper rather than to a grid point short of it.
• step_size must be strictly positive, and must be integer-valued for INT parameters.
• cardinality() accounts for step_size: a grid-valued FLOAT reports the finite number of grid points instead of inf, and a grid-valued INT counts grid points rather than every integer in [lower, upper].

step_size defines a discrete grid but does not, by itself, force discrete acquisition optimization; how the optimizer treats the parameter depends on the grid cardinality and is determined at the generator level.

Differential Revision: D107274057

[meta-codesync]

@saitcakmak has exported this pull request. If you are a Meta employee, you can view the originating Diff in D107274057.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.51%. Comparing base (d0ae700) to head (b575948).

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #5213    +/-   ##
========================================
  Coverage   96.50%   96.51%            
========================================
  Files         617      617            
  Lines       69776    69889   +113     
========================================
+ Hits        67339    67452   +113     
  Misses       2437     2437            

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.