Cardano Updates

Flake lock file updates:

• Updated input 'haskell-nix':
    'github:input-output-hk/haskell.nix/cad2645' (2026-01-03)
  → 'github:input-output-hk/haskell.nix/a1dc767' (2026-02-14)
• Updated input 'haskell-nix/hackage':
    'github:input-output-hk/hackage.nix/05797dd' (2026-01-03)
  → 'github:input-output-hk/hackage.nix/26b99ff' (2026-02-14)
• Updated input 'haskell-nix/hackage-for-stackage':
    'github:input-output-hk/hackage.nix/fba3874' (2026-01-03)
  → 'github:input-output-hk/hackage.nix/0e5a1b2' (2026-02-14)
• Added input 'haskell-nix/hls-2.12':
    'github:haskell/haskell-language-server/7d983de' (2025-09-24)
• Updated input 'haskell-nix/iserv-proxy':
    'github:stable-haskell/iserv-proxy/1506051' (2025-08-15)
  → 'github:stable-haskell/iserv-proxy/91ef7ff' (2026-02-04)
• Updated input 'haskell-nix/stackage':
    'github:input-output-hk/stackage.nix/8a7333a' (2026-01-03)
  → 'github:input-output-hk/stackage.nix/f656c6e' (2026-02-11)

This makes available the functionality that was previously used on the
fly when a package version is forced. Now, it should be done ahead of
time and stored along with the metadata in the git repo, to ensure
future reproducability.

Add --max-retries (default: 7) and --max-concurrent-downloads
(default: 10) flags to `foliage build`.

With 7 retries, curl's exponential backoff spans 1+2+4+8+16+32+64 =
127 seconds, giving GitHub ample time to recover from transient 502s.

Lowering concurrent downloads from 20 to 10 further reduces server
pressure under high parallelism (-j 0).

Add --max-retries (default: 7) and --max-concurrent-downloads
(default: 10) flags to `foliage build`.

With 7 retries, curl's exponential backoff spans 1+2+4+8+16+32+64 =
127 seconds, giving GitHub ample time to recover from transient 502s.

Lowering concurrent downloads from 20 to 10 further reduces server
pressure under high parallelism (-j 0).

Add --max-retries (default: 7) and --max-concurrent-downloads
(default: 10) flags to `foliage build`.

With 7 retries, curl's exponential backoff spans 1+2+4+8+16+32+64 =
127 seconds, giving GitHub ample time to recover from transient 502s.

Lowering concurrent downloads from 20 to 10 further reduces server
pressure under high parallelism (-j 0).

- DeterminateSystems/nix-installer-action: v9 → v21
- DeterminateSystems/magic-nix-cache-action: v2 → v13
- cachix/cachix-action: v14 → v16

Fixes CI failures caused by TypeError: Invalid URL in the old
nix-installer-action when fetching from install.determinate.systems.

Fix transient HTTP 502 failures when downloading source tarballs from
GitHub under high parallelism (-j 0).

Two changes:

1. Add --retry 3 --retry-connrefused to curl invocations. Curl retries
   transient HTTP errors (408, 429, 500, 502, 503, 504) with built-in
   exponential backoff (1s, 2s, 4s). The existing actionRetry 5 is kept
   as a second layer for non-HTTP failures.

2. Cap concurrent downloads at 20 using Shake's Resource mechanism.
   This prevents hundreds of simultaneous curl processes from
   overwhelming GitHub when running with -j 0. Non-network tasks
   (patching, sdist creation, index building) still run at full
   parallelism.

- DeterminateSystems/nix-installer-action: v9 → v21
- DeterminateSystems/magic-nix-cache-action: v2 → v13
- cachix/cachix-action: v14 → v16

Fixes CI failures caused by TypeError: Invalid URL in the old
nix-installer-action when fetching from install.determinate.systems.

Fix transient HTTP 502 failures when downloading source tarballs from
GitHub under high parallelism (-j 0).

Two changes:

1. Add --retry 3 --retry-connrefused to curl invocations. Curl retries
   transient HTTP errors (408, 429, 500, 502, 503, 504) with built-in
   exponential backoff (1s, 2s, 4s). The existing actionRetry 5 is kept
   as a second layer for non-HTTP failures.

2. Cap concurrent downloads at 20 using Shake's Resource mechanism.
   This prevents hundreds of simultaneous curl processes from
   overwhelming GitHub when running with -j 0. Non-network tasks
   (patching, sdist creation, index building) still run at full
   parallelism.