Multiple BeforeAll/BeforeEach/AfterEach/AfterAll in the same container - only the last one runs

[nohwnd]

Found while migrating PowerShell/PowerShell tests to Pester 5 (PR PowerShell/PowerShell#27290). If a container has more than one BeforeAll (or BeforeEach / AfterEach / AfterAll), only the last one runs. The earlier ones are discovered and silently dropped.

Repro on 5.7.1:

Describe "multi-hooks" {
    BeforeAll  { Write-Host "BA1" }
    BeforeAll  { Write-Host "BA2" }
    BeforeEach { Write-Host "  BE1" }
    BeforeEach { Write-Host "  BE2" }
    AfterEach  { Write-Host "  AE1" }
    AfterEach  { Write-Host "  AE2" }
    AfterAll   { Write-Host "AA1" }
    AfterAll   { Write-Host "AA2" }
    It "i" { Write-Host "    It" }
}

Prints BA2, BE2, It, AE2, AA2. The *1 blocks never run.

Mechanism is plain assignment in src/Pester.Runtime.ps1 - New-OneTimeTestSetup does .CurrentBlock.OneTimeTestSetup = , same for New-OneTimeTestTeardown, New-EachTestSetup, New-EachTestTeardown. Last writer wins. The Block type holds a single ScriptBlock per hook, not a list.

This bit a lot of files in the PowerShell repo migration - about -1255 failures across CI once the duplicate blocks were merged. The failure mode is silent: helper functions/variables defined in the earlier block simply aren't there at runtime, and the test fails with CommandNotFoundException or "argument is null" in It.

Idea

Either:

1. Run all blocks of the same hook in source order (append to a list), or
2. Fail discovery with a clear error when more than one is found in the same container.

(2) is the smaller change and would have caught all the cases I just fixed. (1) matches what people coming from xUnit/NUnit-style multiple setup methods would expect, and it's what the current docs imply.