build: bump ModelingToolkitBase for DAE NLStep hook

[ChrisRackauckas-Claude]

Ignore until reviewed by @ChrisRackauckas.

What changed and why

Bump ModelingToolkitBase from 1.62.0 to 1.63.0 so the public generate_DAENLStepData hook added on master can be released. ModelingToolkit master currently permits and resolves registered ModelingToolkitBase 1.62.0 on Julia 1.10, but that release predates the hook and causes ModelingToolkit precompilation to fail.

This is a standalone SemVer-minor release bump. ModelingToolkit already accepts all 1.x ModelingToolkitBase releases through its existing 1.62 compat entry.

Regression evidence

Failing before, from the working-folder parent with clean ModelingToolkit commit d7f3ad523c00b4150281f259d1e97618fba79982, clean LinearSolve commit f5b63ee4ed69a7558d2133a27d73805556d37a16, and Julia 1.10.11:

timeout 3600 julialauncher +1.10 --project=mtk-clean -e 'using Pkg; Pkg.develop(PackageSpec(path=abspath("mtk-ls-clean"))); Pkg.update(); using ModelingToolkit'

ERROR: LoadError: UndefVarError: `generate_DAENLStepData` not defined
@ mtk-clean/src/systems/solver_nlprob.jl:10

Git bisect identified 4ae2e03f00957017e6a28e96196c338f3e944bfa as the first bad commit.

Passing from the ModelingToolkit checkout with the bumped local ModelingToolkitBase release contents:

timeout 3600 julialauncher +1.10 --project=. -e 'using Pkg; Pkg.develop(PackageSpec(path=abspath("lib/ModelingToolkitBase"))); Pkg.develop(PackageSpec(path=abspath("../mtk-ls-clean"))); Pkg.update(); using ModelingToolkit; println("MODELINGTOOLKIT_LOAD_OK")'

MODELINGTOOLKIT_LOAD_OK

Verification

env GROUP=InterfaceI julialauncher +1.10 --project=. -e 'using Pkg; Pkg.test()'

Test Summary: | Pass  Broken  Total      Time
InterfaceI    | 1501       5   1506  24m02.0s
Testing ModelingToolkitBase tests passed

env GROUP=QA julialauncher +1.10 --project=. -e 'using Pkg; Pkg.test()'

JET Tests  | 54 passed, 54 total
Aqua Tests | 10 passed, 1 failed

The Aqua piracy failure reproduces on clean master. Its QA lane has been red since the lane was introduced by 9fadfdca1e0c119ce7d109ae7fa5416ca05ce200. Draft PR 4907 fixes the four JumpProcesses-related methods; the ten historical SymbolicUtils.Code.toexpr methods are tracked by issue 4670. This version-only diff does not touch those methods.

Additional checks:

• git diff --check upstream/master...HEAD: clean.
• typos lib/ModelingToolkitBase/Project.toml: clean.
• Runic is not applicable to a TOML-only diff.

CI status

CI completed with 76 passing, 14 failing, and 3 skipped checks. None of the failures exercise the changed version line:

• Five long jobs reached the two-hour job limit and ended only with The operation was canceled, without a test failure: root QA on Julia 1, root InterfaceII on Julia pre, root Optimization on Julia lts, and ModelingToolkitBase Optimization on Julia lts/pre.
• The two benchmark failures are tracked by issue 4910.
• The Catalyst unexpected passes are fixed on Catalyst master by PR 1512 and await a release.
• The ModelingToolkitBase downgrade, InterfaceI minimum-version, and ModelingToolkitStandardLibrary resolver failures have clean reproductions and focused tracking issues or fixes linked below.
• The remaining Optimization Julia 1/pre error reproduces on clean master and is tracked separately by issue 4925.

Not verified

• The original downstream job cannot resolve 1.63.0 until this bump is merged, tagged, and registered; the explicit local-develop reproducer verifies the release contents in the meantime.
• Docs were not built because this PR only changes package version metadata. The hook documentation and tests landed with the feature commit.
• GPU paths were not run.

Review note

The judgment call is a minor rather than patch bump because generate_DAENLStepData is new public API.

Links

• First bad downstream commit: 4ae2e03
• Existing jump-piracy fix: Avoid piracy in jump dependency traversal #4907
• Existing piracy tracking issue: QA: track pre-existing Aqua/JET/ExplicitImports findings (root + ModelingToolkitBase) #4670
• Benchmark tracking issue: Benchmark script must import SciMLBase.init directly #4910
• Catalyst unexpected-pass fix: Activate corrected conservation-law tests Catalyst.jl#1512
• ModelingToolkitBase DiffEqBase floor issue: Raise ModelingToolkitBase DiffEqBase floor for symbolic diagnostics #4922
• BoundaryValueDiffEqMIRK floor issue: Raise BoundaryValueDiffEqMIRK floor past the RecursiveArrayTools v4 interpolation bug #4923
• LinearSolve floor issue: Raise the LinearSolve 5.x floor past the ForwardDiff issquare import bug #4924
• ModelingToolkitStandardLibrary resolver issue: Test resolver conflict: DataInterpolations 8 vs SciMLBase 3.35+ ModelingToolkitStandardLibrary.jl#498
• ModelingToolkitStandardLibrary resolver fix: Support DataInterpolations 9 ModelingToolkitStandardLibrary.jl#499
• Optimization generated-parameter tracking issue: Optimization tests construct incomplete parameters for expression trajectories #4925
• LinearSolve PR downstream failure: https://github.com/SciML/LinearSolve.jl/actions/runs/31363806735/job/93377835731
• Clean LinearSolve main downstream failure: https://github.com/SciML/LinearSolve.jl/actions/runs/31361291552/job/93370530272

[AayushSabharwal]

Already done