SCurve: fix calculate_path bugs and add comprehensive tests

[lthall]

Summary

Fix three bugs in SCurve::calculate_path and set_speed_max, add tests
that cover all code paths, and clean up float literal consistency.

Fixes #32724

Bug fixes:

• Fix sign error in path length acceleration constraint: (L + 4V0tj)
  should be (L - 4V0tj). The old formula inflated the allowed
  acceleration when V0 > 0, producing invalid motion profiles.
• Add early return when At <= 0. After the sign fix, At can go negative
  when the path is too short for the initial velocity.
• Fix integer division in set_speed_max: powf(x, 1/3) evaluated the
  exponent as 0, making powf always return 1.0 instead of the cube
  root. This disabled the velocity-based jerk time limit.
• Guard against divide-by-zero where dP is divided by
  segment[SEG_CONST].end_vel in four places.

Tests:

• 32 test cases covering all 8 code paths in calculate_path with 4
  inputs per path near transitions and midpoints.
• Constraint validation replicating the add_segments path structure to
  verify total distance, final velocity, final acceleration, and that
  velocity, acceleration, and jerk limits are not exceeded.

Cleanup:

• Replace double literals with float literals throughout to avoid
  implicit double-to-float conversions.
• Fix misleading comment, whitespace, and operator spacing.

Board,antennatracker,blimp,bootloader,copter,heli,plane,rover,sub
CubeOrange,*,*,*,128,128,136,136,128

Classification & Testing (check all that apply and add your own)

• Checked by a human programmer
• Non-functional change
• No-binary change
• Infrastructure change (e.g. unit tests, helper scripts)
• Automated test(s) verify changes (e.g. unit test, autotest)
• Tested manually, description below (e.g. SITL)
• Tested on hardware
• Logs attached
• Logs available on request

[amilcarlucas]

Binary Name      Text [B]        Data [B]     BSS (B)        Total Flash Change [B] (%)      Flash Free After PR (B)
---------------  --------------  -----------  -------------  ----------------------------  -------------------------
blimp            0 (0.0000%)     0 (0.0000%)  0 (0.0000%)    0 (0.0000%)                                      780972
arducopter       128 (+0.0081%)  0 (0.0000%)  0 (0.0000%)    128 (+0.0080%)                                   374116
ardusub          124 (+0.0087%)  0 (0.0000%)  -4 (-0.0015%)  124 (+0.0086%)                                   530428
arducopter-heli  128 (+0.0081%)  0 (0.0000%)  0 (0.0000%)    128 (+0.0081%)                                   378508
arduplane        124 (+0.0078%)  0 (0.0000%)  -4 (-0.0016%)  124 (+0.0078%)                                   378184
ardurover        124 (+0.0086%)  0 (0.0000%)  4 (+0.0015%)   124 (+0.0085%)                                   512868
antennatracker   0 (0.0000%)     0 (0.0000%)  0 (0.0000%)    0 (0.0000%)                                      800056

[tpwrules]

Is this observable in flight?

[rmackay9]

Hi @tpwrules, my understanding from a quick chat with Leonard is that it is observable as an overshoot as the vehicle transitions from one waypoint to the next.

[lthall]

Is this observable in flight?

I have tired to excite this but it is really difficult. I believe I have seen it once but then could not replicate it. I have spent some time trying but have not been succsessful.

[andyp1per]

It's 124 bytes for doing the same check multiple times. You could probably avoid the impact of this by using a function

[lthall]

Good suggestion. Addressed in a follow up PR:

#32787

[Georacer]

This has been included in 4.7.0-beta4. Thanks!