HEFT: reproduce a classic result

Implement HEFT and reproduce the makespan of 80 from the original paper's 10-task example.

Input: the Chapter 6 format (see OS6.1).

Algorithm:

1. Compute rank_u (as in OS6.1). Priority order: decreasing rank_u; ties → smaller task number.
2. For each task in that order and each processor k (in order P1, P2, …):
   • ready time = max over predecessors q of (finish(q) + (0 if q runs on k else c(q, i))), or 0 for entry tasks;
   • EST = the earliest start ≥ ready time such that the interval [EST, EST + w(i, k)) fits in an idle gap of k (insertion policy: gaps between already scheduled tasks count; otherwise after the last task);
   • EFT = EST + w(i, k). Choose the processor with the minimum EFT (ties → lowest processor number).

Output:

Upward ranks: T1=... T2=... (two decimals, task order)
Priority order: T1 T3 ...
T1 -> P3 start 0 finish 9        (one line per task, in scheduling order)
...
Makespan: 80
SLR: 1.951 (CP_MIN = 41)
Speedup: 1.588 (best sequential = 127)
Efficiency: 0.529

Times are printed as integers when they are whole numbers, otherwise with two decimals. CP_MIN = the longest path using each task's minimum cost and no communication; best sequential = the minimum over processors of the sum of all task costs on that processor.

The example input is the classic one (10 tasks, 3 processors) — see the test data in the model solution.