The StandardErrorListener used AnsiConsole.out directly to print
progress messages, bypassing Nextflow's logger which respects the
`-q` flag. Add a `quiet` parameter to StandardErrorListener that
suppresses progress output (beforeAll, beforeFile, beforeFormat,
afterAll) while still showing errors and warnings.

Signed-off-by: nextflow-io <info@nextflow.io>

https://claude.ai/code/session_01MY4FcPPaSmACrBsdpHTfnT
Signed-off-by: Claude <noreply@anthropic.com>

Two issues were causing test failures:

1. StandardErrorListener was printing via AnsiConsole.out, which is a
   static field initialized at class-load time and not updated when
   OutputCapture replaces System.out. Switch all output to System.out
   so test capture works correctly.

2. CmdLint is @CompileStatic, so the Groovy compiler generates a call
   to isQuiet() (boolean convention) rather than getQuiet() when
   accessing launcher.options.quiet. The test stubs were using
   getQuiet() >> true, which was never called, so quiet was always
   false. Fix by stubbing isQuiet() >> true instead.

Signed-off-by: Phil Ewels <phil.ewels@seqera.io>

Introduce a private print() helper that consolidates all output via a
single code path:

- When ansiLog=true, delegates to AnsiConsole.out, which handles TTY
  detection and strips ANSI escape codes when writing to a pipe.
- When ansiLog=false, sets Ansi.setEnabled(false) immediately before
  rendering the Ansi object to plain text, then writes to System.out.
  Setting the flag just before toString() ensures the global Ansi state
  is correct at render time regardless of what other components may have
  set in between. Writing to System.out (rather than the static
  AnsiConsole.out field) means the output is captured correctly by test
  infrastructure that replaces System.out.

Signed-off-by: Phil Ewels <phil.ewels@seqera.io>

Ansi.setEnabled() is static-only in Jansi 2.x -- there is no per-instance
flag. Ansi.toString() returns the raw StringBuilder contents; escape codes
are baked in when fg()/bold() etc. are called, not at render time.

The previous approach of calling Ansi.setEnabled(false) in print() just
before toString() was too late: codes were already in the buffer.

Fix: call Ansi.setEnabled(ansiLog) inside the ansi() helper, before every
Ansi object is created. This ensures the global flag is correct for all
subsequent append operations, so plain-text output is produced when
ansiLog=false (e.g. when writing to a pipe in integration tests).

Signed-off-by: Phil Ewels <phil.ewels@seqera.io>